EIB Papers Volume 01. n°1-1996
-
Upload
khangminh22 -
Category
Documents
-
view
2 -
download
0
Transcript of EIB Papers Volume 01. n°1-1996
Special Issue on Emu
The steeple chase towards EMU Daniel Gros
On the feasibility of EMU with diverse European economic transmission mechanisms
Ole Rummel
15
51
Is Maastricht α good contract? Bernhard Winkler
How to fix conversion rates at the start of EMU Paul De Grauwe
A proposal to stabilise the value of the ECU Luis Gonzalez-Pacheco & Alfred Steinherr
Book review H. W. J. Bosnian
79
97
119
139
I 4
•-C/5
( ^ V,
BANQUE EUROPÉENNE D'INVESTISSEMENT EUROPEAN INVESTMENT BANK
Hdîtonal policy
The EIB Papers are published tv^ice α year by the Chief Economist's Department of the
European Investment Bank. The journal is aimed at encouraging high-quality economic
research and debate on matters of α European interest. As such the Papers are intended
to be accessible to non-specialist readers and emphasise policy dimensions rather than
technical issues. They present the results of research carried out by Bank staff together
with contributions from external scholars and specialists. The winning essays of the biennial
EIB Prize are also published in the winter edition during competition years.
Articles will only be accepted for publication on the condition that they have not already
been published elsewhere. All articles in the EIB Papers may be freely reproduced and
quoted; however, the Editor would appreciate an appropriate acknowledgement and α
copy of the publication in question.
The views expressed in the articles are those of the individual authors and do not
necessarily reflect the position of the EIB.
Advisory Board
Alfred Steinherr (Chairman)
Pier-Luigi Gi l ibert, Jacques G i r a r d , Theohorry Grammatikos,
Daniel Ottolenghi, Eric Perée
E d i t o r
Christopher Hurst
Les Cahiers de la BEI sont publiés deux fois par an par le département de l'économiste
en chef de la Banque Européenne d'Investissement. Le journal α pour but d'encourager
la recherche de haute qualité en économie et le débat sur des sujets d'intérêt européen.
En tant que tels, il est attendu des Cahiers qu'ils soient accessibles aux lecteurs non
spécialistes et qu'ils mettent l'accent sur les dimensions politiques plutôt que les problèmes
techniques. Ils présentent les résultats de la recherche menée par le staff de la Banque
ainsi que des contributions de chercheurs et de spécialistes extérieurs. Les essais
gagnants du Prix de la BEI proposé tous les deux ans sont aussi publiés dans l'édition
d'hiver les années de compétition.
Les articles ne seront acceptés pour publication qu'à condition qu'ils n'aient été publiés
ailleurs. Tous les articles des Cahiers de la BEI peuvent être librement reproduits et
cités; cependant, l'Editeur apprécierait qu'un crédit approprié leur soit donné et qu'une
copie de la publication lui soit envoyé.
Les vues exprimées dans les articles sont personnelles aux auteurs et ne reflètent pas
nécessairement la position de la BEI.
Contents
5 Preface by Wolfgang Roth, Vice President
Special issue o n EMU
7 Editor's Introduction
Daniel Gros
Ole Rummel
15 The steeple chose towards EMU
51 On the feasibility of EMU with diverse European
economic transmission mechanisms
Bernhard Winkler 79 Is Maastricht α good contract ?
Paul De Grauwe 9 7 How to fix conversion rotes at the start of EMU
Luis Gonzalez-Pacheco & 119 A proposal to stabilise the value of the ECU
Alfred Steinherr
H.W.J. Bosmon 139 Book review
ί :B Pape-s Volume 1 Noi 1996
Preface
mf
For a decade the EIB Papers has published the occasional research of EIB staff
:** memt)ers. In a recent review of communications policy, it was decided that it
^ would be desirable to present a more structured EIB contribution to the EU poli-
Wolfqanq Roth ^Y debate. It is thus with pleasure that I now introduce you to the new version of
Vice-President ' ' 'e Papers.
In the future, the Papers wi l l be published twice a year by the Chief Economist's
Department. The journal wi l l be used to publish high-quality economic analysis
on a b road range of European issues. It wi l l include the results of research car
r ied out by Bank staff together with contributions from external scholars and spe
cialists. As the goa l is to stimulate debate in the broader community, the articles
in the Papers wi l l emphasise pol icy dimensions rather than technical issues.
It is intended that many of the editions wi l l focus on a special topic. In this first
issue of the new series, we develop some themes related to European Monetary
Union. N o single event wi l l change our lives more profoundly than the introduc
tion of the future single currency. As the Union's financing institution, the EIB has
been furthering European integration for almost 4 0 years. W e expect the road
to an EU-wide monetary union, including future new member states, wi l l pose
special challenges to al l f inancial institutions. However, we see EMU as only
one further — and far from the final — step in the process of European economic
integration.
The Papers wi l l also be used to publish the winning essays of the biennial EIB
Prize. In 1983, on the occasion of its 25th Anniversary, the Bank established a
prize to be awarded every two years for a doctoral dissertation on a topic rela
ted to investment and finance. After more than ten years without a change, we
have decided it is also timely to take a new direction here. As from 1997, the
EIB Prize wi l l be awarded for short essays, with the goa l of stimulating new
work and providing an incentive for wider research on European themes.
Details of the prize, including entry conditions, are given on the next page.
I encourage you to enter the competition.
[ IB Popes
Volume 1 Noi 1996
1997 EIB PRIZE
The European Investment Bank invites entries for the 1 9 9 7 EIB Prize, which consists of:
ECU 10 0 0 0 (first prize)
ECU 7 5 0 0 (second prize)
ECU 5 0 0 0 (third prize)
ECU 5 0 0 0 (special topic)
and three ECU 1 0 0 0 prizes for entries from persons under the age of 30.
The EIB Prize is offered for short essays on economic and financial topics related to
European affairs.
In addition, there is α special award for essays on the
1997 set topic: "In economic terms, has the natioivstate
become obsolete in the European Union?".
To qualify, all essays must be written in α non-technical
style, be of α maximum length of 5 0 0 0 words and must
be unpublished in the form presented.
Entries must be submitted before 1 March 1997.
The EIB Prize is open to any person under 45 years of
age who has the nationality of α Member State of the
EU, or α member state of EFTA, or α Europe Agreement
country.
Winning entries are selected by α Prize Jury: Lord Roll
of Ipsden (Chairman), Antonio Borges, Edmond
Malinvaud, Alberto Quadrio-Curzio, Helmut
Schlesinger, Jacques-François Thisse, and Alfred
Steinherr, Chief Economist of the EIB.
The prizes will be presented at α conference to be held in Florence in October 1997
and the EIB will publish all winning entries.
The EIB Prize Rules may be obtained from:
The Chief Economist,
European Investment Bank,
L-2950 Luxembourg
Tel : (352) 4379-3438, Fax : (352) 4379-3492;
EUROPEAN INVESTMENT BANK
E IB Papers
Volume! Noi 1996
Editor's introduction
Hardly a day seems to pass without a new
article being published on EMU, and so
to choose this as the special topic for the
first edition of the new EIB Papers may
not appear particularly original. However,
the launch of EMU remains the single most
important issue on the European econ
omic policy agenda, and there are two
broad themes where we believe additional
discussion is merited:
• How should we interpret the convergence
criteria?
• How do we manage foreign exchange
markets between now and the irrevo
cable fixing of exchange rates?
For an overview of the issues we turned
to a major contributor to the EMU debate,
Daniel Gros of the Centre for European
Policy Studies.
Of all the Maastricht convergence criteria,
it is probably the management of public
finances where there are the most import
ant outstanding issues, and it is here that
Gros focuses much of his discussion.
Since some countries had debt levels well
in excess of the target limit at the time of
signing the Treaty, the wording of this
condition is very loose - only that debt
should be falling towards the reference
level (of 60 percent of GDP] at a satisfac
tory pace. If this condition is interpreted
in a very flexible manner to help highly
indebted countries, it could hurt the credi
bility of EMU and make it seem that
some countries were being treated more
leniently than others.
Gros notes that the figures in the Treaty
are more or less consistent with each
other and a reasonable economic
performance. A deficit pushes the debt
ratio up, while growth lowers this ratio
by increasing the denominator. If how
ever, nominal GDP growth were 5 percent
(say inflation of 2 percent and real growth
of 3 percent) and the fiscal deficit were
3 percent, then the debt ratio would con
verge to 60 percent over the long-term
(equal to the deficit divided by the nominal
growth rate). Moreover, the difference
between the present debt level and the
60 percent target is reduced at by a rate
equal to nominal growth (or 5 percent per
annum in this casej.
Gros takes this observation one step fur
ther. He proposes that if excessive debt,
on average, falls towards the target at a
rate of at least 5 percent per annum, this
should be considered as "satisfactory" in
terms of the Treaty. However, in some
Volume 1 Noi 1996
countries off-budget debt accumulation is
continuing at some 1-2 percent of GDP
per annum. Here, the permissible deficit
(of 3 percent of GDP] would have to be
reduced by an equivalent amount if total
debt were to decrease as desired.
The problem with this approach is that
the figure of 5 percent remains some
what arbitrary. The 3 percent fiscal deficit
is an upper limit, and if the average deficit
were less (say 1.5 percent) a long-term
equilibrium debt ratio of 6 0 percent
would no longer be consistent with a
nominal growth of 5 percent (debt would
converge to only 3 0 percent of GDP).
Equally, fhe future nominal growth may
well be less than 5 percent - especially if
the future European Central Bank (ECB) is
successful at keeping low inflation rates.
However, the observation that at any one
time only one of the criteria (either the
deficit or debt reduction) can be driving
fiscal policy is key. One could take Gros '
logic to its extreme and argue that if the
deficit is no more than 3 percent, then
debt will, by definition, be diminishing at
a satisfactory pace. In such a situation,
the debt ratio should fall over the coming
years ( I), and it is not evident why a
greater rate of decline would be needed
to establish credibility on financial mar
kets (consider the case of Belgium). A
debt ratio criteria would still be needed,
but it would serve another purpose - to
limit the possibility of government running
up excessive off-budget debt that would
be converted into a liability at some future
date.
After discussing a range of related issues.
Gros concludes with a proposal for the
management of exchange rates for those
countries that do not enter EMU in the
first wave. Since reduced interest rates
lower debt sen/ice, and this is a major
component of government expenditure, it
is possible to get into a virtuous spiral:
confidence that a country will enter EMU
lowers interest rates (this has been very
marked during 1996 for Italy and Spain
(2)); this makes it easier to meet the entry
requirements; confidence increases ...
Conversely, countries that are thought to
be making an extra convergence effort
due only fo Maastricht can expect a
strong reaction from financial markets if
they do not enter in January, 1999.
Can anything be done to limit this risk?
Gros proposes a temporary "Associate
Membership" of EMU. Here, countries
would accept all the obligations arising
from EMU membership, but would not
participate in the ECB decision-making
process. This should be credible if the
proposal is seen as a "currency board"
and the national Central Bank possesses
adequate foreign currency reserves to
guarantee conversion of all of its liabilities.
This would be the case for most candidate
/ j Strictly speaking, the debt ratio would only be falling i f the deficit was less than the current debt ratio times
nominal growth. If the debt ratio were 120 percent, a deficit of 3 percent would require nominal growth of
more than 2.5 percent to keep debt on a d o w n w a r d track. With an average deficit of 1.5 percent, growth of
just over 1 percent would suffice.
2] However, even within EMU more highly indebted countries can expect to p a y a risk premium over the
more creditworthy members. An example is given by the Provinces of Canada. Here the highest-rated
Province (AA+j pays 5 0 bp less than the weaker-rated (BBB+j.
ί Ή Paoo's
Volume 1 N o i 1 9 9 6
countries. Credibility would be greatly en
hanced if the country followed full EMU
members in converting its public debt to
Euro (if debt is Euro-denominated the
cost of devaluing would be very high).
Clearly, a government could "go it alone"
with a currency board, indeed, it is not
possible to exclude any country from shar
ing in EMU if it is ready to forego partici
pation in ECB management. However, a
formal agreement between the ECB and
the country concerned could only further
enhance credibility.
The Maastricht convergence criteria have
been disturbing for some economists
since they do not seem to be consistent
with the normal requirements for an
"optimal currency area ". These relate to
microeconomic structures rather than
macroeconomic conditions. For example,
an important issue is the flexibility of la
bour markets. This permits a region to ad
just to an external shock, not hitting other
members of the monetary union, without
long periods of unemployment. Equally, if
there is no such thing as an asymmetric
shock, in the sense that all countries are
hit by identical economic shocks, then an
independent monetary policy would be
of limited value. The literature on opti
mal currency areas is large, but the
general conclusion is that EMU would be
appropriate for a core group of European
countries, but may not be ideal for the
EU-15 as a whole.
Ole Rummel (EIB) elaborates on this dis
cussion. He notes that beyond the ques
tion ofthe "symmetry" of an external shock.
the macroeconomic effect also depends
upon how these shocks are transmitted
through the various European economies.
An example of possible differences is
given by the prevalence of short-term per
sonal borrowing and floating-rate mort
gages in the UK. This could make the UK
much more sensitive to changes in short-
term interest rates than its continental part
ners. Similar differences could apply for
a range of other macroeconomic variables.
Rummel estimates (3) the short-term linkages
between a range of economic variables,
and studies how European shocks to the
money supply, long-term interest rates,
inflation, and industrial production pro
pagate through ten member states. There
are clear differences between countries.
The initial impact of a shock is varied,
and the speed at which this impact dies
away is also very different. Only Germany
and the Benelux consistently act in a
similar way. In the extreme, only these
would be in the EMU core.
However, this analysis does not assess
whether the costs that could arise in a
country with atypical transmission channels
would be greater than the benefits of
membership. It may also be that many
facets of an optimal currency area are
endogenous. That is to say that economic
integration following EMU, such as streng
thening trade links and closer correlation
of business cycles, means that countries
are much more likely to meet the require
ments for an optimal currency area once
they have actually entered into it. A similar
logic could apply to the transmission
mechanisms analysed by Rummel. The
3} With a Bayesian Vector Autoregression (VARj methodology.
E 'B PQDCS
Volume 1 N o i 1996
important point is that successful member
ship of EMU may require structural
reforms, such as liberalisation of labour
markets, to complement the "natural"
integration process.
This brings us back to the macroeconomic
criteria for EMU membership actually set
out in the Treaty. An alternative view is that
reaching these entry conditions should be
seen as a sort of noble and virtuous deed
niat permits membership of the ECB "round
table". Bernhard Winkler (European
University Institute, Florence (4jj examines
this somewhat frivolous description in a
more structured game theoretic framework.
As a simplification, Europe can be split
into two groups: one where there is
already a high monetary credibility and
which is concerned about loss of reputa
tion and price stability under EMU. These
countries prefer convergence and credi
bility to be established prior to member
ship of EMU. As far as the second group
is concerned, EMU offers the possibility
of obtaining credibility through entry. For
this group convergence should come
after membership, if at all (5). There are a
range of possible outcomes to this game (6).
EMU without convergence would be bad
for high credibility countries, while
convergence without EMU would be a
neg-ative outcome for low credibility
countries (assuming there is additional
Maastricht induced convergence above
the level which a country would find in
its own interest to perform). The result is
that countries get stuck in an equilibrium
where nothing happens, neither con
vergence nor EMU. This is the well-known
"Prisoners' Dilemma", where the players
of a game may be led by self-interest to
take decisions that are mutually disad
vantageous (7j.
Winkler discusses how the Maastricht
Treaty can been seen as a way of ensur
ing the optimal co-operative outcome of
both convergence and EMU. The approach
of setting a fixed-date when EMU will
happen January 1999], coupled with
qualified majority voting for entry decisions
sets a framework where both groups can
commit to an EMU strategy. The risk of
re-negotiation of the contract immediately
prior to entry may be limited through third-
party arbitration - perhaps one role for
the convergence reports of the Commission
and the European Monetary Institute.
Since fhe benefits of joining EMU de
pend upon the number of countries that
go in, an individual country's convergence
efforts also benefit other members (thus
membership has the features of a public
good]. Again, fhe setting of a fixed-date
for EMU is a way of limiting this co-ordi
nation problem. If also explains why some
countries appear fo have started con-
4) Bernhard Winkler was awarded the EIB's Campilli-Formentini Scholarship to finance his research at fhe
European University institute.
5 ] The point is that the rules of the club wilt be interpreted by its members. Thus, the goa l of price stability is
not fully credible i f profligate countries enter. Equally, members may be forced to "bait-out" weaker partners,
if only with grant payments.
6) Clearly, this excludes countries which exercise and opt-out, and so choose not to p lay the game.
7} Two prisoners are brought in and interrogated separately. Each knows they wil l both get off if neither
confesses. However, they are both told that if only one talks, the other wi l l receive a particularly heavy sen
tence. In this case, both players may decide to protect themselves by confessing.
10
E iB Popes
Volume 1 N o i 1996
vergence late in the day. If was rational
to wait and see what happened as long
as other countries did the same.
Winkler's analysis presents fhe Maastricht
Treaty as a contract that moves partici
pants to a socially optimal equilibrium
that would not occur of its own accord.
This framework also highlights the risk of
coun-tries that do not enter In the first
round being permanently left out unless a
new framework is in place that maintains
sufficient incentives for convergence. The
importance of some kind of post-1999
agreement, such as the "Associate Mem
bership" of EMU proposed by Gros, is
once more apparent.
Speaking a few years ago at a conference
in Luxemtmurg in honour of Pierre Werner,
(8) Sir Edward Heath mused: "I always vis
ualise our Ministers of Finance, meeting
quietly one Saturday at five to hvelve mid
night. They agree "we are going to have
a single currency at existing rates ". And at
five past twelve they tell the world "Europe
has a single currency. " Unfortunately, rea
lity will not be so simple. The second
theme of this edition discusses how currency
fluctuations could be managed until ex
change rates are irrevocably fixed.
When the future core members of EMU
are decided, the authorities will have to
make some announcement of how future
conversion rates will be set This could
either be a pre-specified fixed-rate, or an
agreed principle of some kind. An example
of the latter would be the "tamfalussy
rule" (9) whereby the conversion rate will
be an average of the market rates for a
given period of time before I January,
1999 (assuming EMU goes ahead at that
date]. Paul De Grauwe (University of
Leuven] had been analysing this ques
tion, and we asked him to explain the
pros and cons of the different approaches.
De Grauwe shows that the tamfalussy
rule produces some surprising results. The
problem arises because the forecast
evolution of the moving average of the
exchange rate becomes a factor that
determines future movements. Every time
some news arrives that changes the ex
change rate, market agents also change
their expectations for all future periods.
This changes the forecast moving average.
In turn, this feeds back into today's expec
tations. De Grauwe likens this to an echo
that bounces back from a wall at the end
of the averaging period. The result is a
sudden increase in the volatility of exchange
rates in the short term, and a possible drift
away the exchange rate that existed
when the conversion rule was announced.
De Grauwe considers applying the
tamfalussy rule retro-actively (e.g. the
future rate will be an average of the past
(1996] and the future (1997 and 1998]].
Unfortunately, this does not solve the pro
blem especially if the authorities are un
fortunate enough to announce the rule
when the trend in the exchange rate and
the trend in the moving average are
going in opposite directions. In this case,
8) The proceeding of the conference are published in: A. Steinherr fed] (19931, 3 0 years of European mone
tary integration: From the Werner Plan to EMU, Longman: London.
9) Scxal led because it has been advocated by Alexandre Lamfalussy, the President of the European
Monetary Institute.
E IB Papers
Volume 1 N o i 1996 11
there can be very large jumps in the
exchange rate at the time of announce
ment. De Grauwe concludes that the
Lamfalussy rule produces more problems
than it solves.
The question becomes one of credibility.
One could argue that the Lamfalussy rule
is inherently more credible than announc
ing a fixed rate exactly because it allows
the exchange rate to vary as the funda
mentals change. This has a cost of gener
ating higher volatility for an initial period,
but this could be partially reduced by
heavy front-loading of the formula (e.g.
using weightings of 60 percent for 1996,
3 0 percent for 1997, and 10 percent for
1998]. As front-loading increases this
becomes progressively like announcing a
fixed-rate anyway. The question is whether
there is some point where credibility is
maximised.
De Grauwe argues for a different solution.
With an adequate commitment at the
moment of announcement, a fixed ex
change rate can also be credible. Such
a commitment could be an agreement
that conversion rates cannot be changed
except by unanimity, coupled with "insti
tutional" front-loading. This would require
that countries would put into practice
some institutional changes that would
normally only occur on 1 January, 1999.
Thus, monetary policies would be decided
jointly in 1998, and each participating
Central Bank would supply its own money
in unlimited amounts in exchange for a
currency under pressure. We arrive again
at the concept of some sort of "Associate
EMU Membership", but with different
rules from those proposed by Gros. The
main differences pertain to joint decision
making, since the relationship now re
flects a marriage of equals, rather than the
wooing of a rich bride by a poor groom.
Luis Gonzalez-Pacheco and Al f red
Steinherr (EIB] also ask whether there are
policy decisions to be taken in managing
exchange rates in the run-up to EMU. In
particular, they look at the value of the
ECU vis-à-vis its constituent elements. The
private ECU is unlike other currencies in
that its value is not established by any
monetary authority. Prior to 1988, a
group of major European banks (the ECU
clearing banks] accepted to convert the
private ECU into the basket of its compo
nents. This ensured that the private ECU
remained very close to its official value.
In 1988, this system broke down as the
banks were no longer prepared to accept
the daily exchange rate risks this entailed
nor the transaction costs of bundling and
un-bundling the basket.
The result was the creation of the ECU
"delta " - a gap between the market value
of the private ECU and the basket ECU. (9]
At times the delta has been quite large,
reaching a peak of 3 percent in early
1996. During 1996 increased confidence
that EMU will go ahead as planned has
much reduced the discount on private
ECUs, since they will be converted to
Euro at a one-for-one basis. This defines
a future value for the ECU.
Unfortunately, it is possible that optimism
for EMU will dwindle and that there will
be renewed periods of instability on
12 ί:1Β Papers
Volume! N o i 1 9 9 6
foreign exchange markets before 1999.
Gonzalez-Pacheco and Steinherr examine
the mechanism that determines the value
of the private ECU. In essence, the size
of the delta results from the expected d i f
ferential between the private ECU interest
rate (given by the intervention rate of the
ECU clearing system] and those of the
basket currencies.
Using an efficient market hypothesis (i.e.
assuming that interest rates and exchange
rates follow a random walk], Gonzalez-
Pacheco and Steinherr analyse the likely
evolution of the ECU delta under the cur
rent arrangement. They show that quite
large fluctuations are possible, and that
once a delta appears one cannot rely on
it disappearing automatically over time. A
simple solution is to guarantee conversion
between the private ECU and the basket
ECU at only certain (though regular]
dates. Gonzalez-Pacheco and Steinherr
show how this limits the maximum delta
and ensures that the expected long-term
value of the delta is zero (i.e. permanent
drift is not possible]. This should be possible
wi&t minimum cast to market participants.
Since a new currency crisis could result
in a rapid widening of the delta (and a
loss of confidence in the ECU], this is an
important point to examine further Though
of a lesser dimension, it fits in with the
development of institutional structures
advocated by Gros and De Grauwe.
Many points have been analysed in this
issue of the Papers. To conclude, some
general points are worth reiterating:
• Assessing appropriate criteria for
membership of EMU is a complex matter
Critics have argued that meeting the
convergence criteria is little more than
the "hazing" of new recruits. But this
does not make the criteria unnecessary
or any less valid.
• Some of the Maastricht conditions are
vague. It would be useful if fhe way in
which these conditions will be interpreted
could be more clearly agreed in advance.
However, the recent discussion over the
post-EMU Stability Pact shows this may
be far from easy.
• The goal is credibility, both for choosing
members and for managing foreign ex
change markets in the interim. One ap
proach is to put more effort into institu
tion building, both ahead of EMU, and
afterwards for countries not in the core.
• At the microeconomic level, adapting
national institutional structures will require
another look at labour markets.
In the absence of hard and fast rules for
creating a well-functioning monetary
union, more effort in early institution buil
ding may be needed. Indeed, this may
be the only way to limit the creation of
long lasting divisions in Europe.
Christopher Hurst
Chief Economist's Department
10) In most cases the private ECU has been worth less than the basket ECU, though the reverse occurred for
a per iod in early 1990.
Ì iB Poce-s
Volume 1 N o i 1996 13
The steeple chose tovs^ords EMU
Daniel Gros
Senior Research Fellow, CEPS, Brussels
1 . I n t r o d u c t i o n
The year 1996 has seen α transformation of the prospect for EMU. Until the middle of
this year it seemed that EMU was α remote prospect and that the preparations for EMU
were "as quaint and as potent as α rain dance" to quote α prominent European politi
cian.
w h a t has prompted this revival of the prospects for EMU? The dogged determination of
politicians is one explanation. But it is likely that this determination was not just motivated
by political considerations. Another driving factor must hove been the realization that the
economic benefits of EMU are perhaps more substantial than had been assumed so for.
Until recently discussions about the economics of EMU were dominated by the optimum-
currency-area approach which suggests that differences in economic structure should be
α major determinant of the cost of EMU. This has often been taken to imply that only a
small group of countries around France and Germany would benefit from EMU.
However, recent studies (see Gros (1996α) for more references) suggest that this might
be the case for α larger area. More importantly, the optimum-currency-area criteria have
turned out to be only of marginal relevance because the central thesis of this approach,
namely that shocks to trade could lead to large unemployment problems, cannot be
confirmed empirically. Recent experience with large exchange rote movements within
Europe shows that exchange rotes do not constitute α powerful tool with which to correct
domestic macroeconomic disequilibrio. On the contrary they seem to be α source of
uncertainty that has α strong negative impact on the economy.
The key problem of the optimum-currency-area approach is thus that it implicitly com
pares the fixing of exchange rates to on idealised world where everything else is
unchanged but exchange rotes con be moved in an ideal manner in response to asym
metric shocks. In reality, the choice might be quite different. In the current economic and
political context, the choice for Europe is, in fact, between the following two alternatives:
i) Erratically moving flexible rates for all member countries (except α narrow DM-bloc)
with the associated dangers for the single market; or
ii) An EMU that expands quickly as the remaining countries satisfy the convergence cri
teria. This would preserve and even strengthen the single market.
This contribution is based on CEPS Paper 65. I wish to thank members of the CEPS Economic Policy Group for
their permission to use our joint work.
t ß Pope's
Volume 1 Noi 1996 15
The n e t bene f i t s f r o m
EMU v^ou ld a p p e a r to
b e r a t h e r l a r g e ; l a r g e r
tììan be fore 1992 ,
w h e n one cou ld assume
tha t the a l ternat ive v^as
a s m o o t h l y f unc t i on ing
EMS.
From this perspective, the net benefits from EMU would appear to be rather large; larger
than what they were thought to be before 1 992, when one could assume that the alter
native to EMU was α smoothly functioning EMS which posed no threats to the single mar
ket. EMU would not only preserve the single market, it should also have some direct
benefits by eliminating excess exchange rote variability. Gros (1996c) shows the very
high exchange rate variability of 1 995 might have increased unemployment in Germany
by about one full percentage point.
2 . The r e m a i n i n g h u r d l e s
The economic benefits of EMU are thus better appreciated today, but there remain obstacles
in the form of the convergence criteria - and for some countries in the form of political will.
This section concentrates on the 11 member countries that hove shown the political will to
participate in EMU and have α chance to do so. The four 'definite outs' are: the UK and
Denmark (which ore most likely to use their opt-out), Sweden which will self-disqualify itself
by not participating in the ERM and Greece, which, by common consent, is too far from
meeting any of the convergence criteria to have α chance to participate in EMU by 1999.
As the performance of these 1 1 effective candidate countries will be closely scrutinized
by hnanciol markets during all of 1 997 this section will discuss some of the convergence
criteria. Table 1 presents the relevant dato as of end-1996. The examination in 1998
will, of course, be based on the definite data for 1997 that will become available in the
spring of that year. The forecasts for 1 997 that ore available now are discussed below
and their likely accuracy is briefly discussed in the annex.
Table 1. Criteria for EMU membership as of late 1996
Β
D
E
F
IRL
1
NL
A
Ρ
FIN
1
Inflation (1)
OK
OK
N O
OK
OK
N O
OK
OK
N O
OK
OK
Interest rate (2)
OK
OK
OK
OK
OK
OK
OK
OK
OK
OK
OK
Excessive
deficit (3)
N O
N O
N O
N O
OK
N O
N O
N O
N O
N O
OK
ERM
membership
OK
OK
OK
OK
OK
N O
OK
OK
OK
OK
OK
Independent
central bank
OK
OK
OK
OK
OK
OK
OK
OK
OK
OK
OK
16
EIB Papers
Volume 1 Noi 1996
Non-candidates:
Inflation (1) Interest rote (2) Excessive ERM Independent
deficit (3) membership central bonk
DK
UK
S
GR
OK
OK
OK
NO
OK
OK
OK
NO
OK
NO
NO
NO
OK
NO
NO
NO
OK
NO
OK
OK
Source: European Commission.
1] At most 1.5 % above the average of the three best performers.
2) At most 2 % above the average of the three best performers in terms of inflation.
3) Existence of an excessive deficits: The entry 'No' means that ECOFIN has found the country to have on excessive deficit in the sense of Article 104c.
There are three criteria that do not need to be discussed here in detail; i) Independence
of the national central bonk is implicitly also α criterion, but it is easy to fulfil, provided
the country has the political will to enact the necessary legislation. All member countries
except the UK hove done so; ii) The criterion concerning interest rates (a long term inter
est rate at most 200 basis points above that of the three best performers in terms of infla
tion) is now fulfilled by all candidates; and iii) The criterion concerning exchange rate
stability (membership in the ERM with 'normal' bands of fluctuations) is formally fulfilled
automatically by all countries that ore members by the ERM as of end 1996. This
includes all member countries with the political will to participate in the first wave of
EMU (except possibly Italy).
It is often argued that these latter two criteria contain self-fulfilling prophecies in the sense
that exchange rotes will be stable and interest rotes low if the country concerned has α
good chance to participate in EMU. But, so the argument goes, if financial markets an
ticipate that the country will not be allowed to participate in EMU exchange rates will
come under pressure and interest rotes will rise. A mechanism of this sort might hove
operated in 1995 (see Gros (1996b) and Obstfeld (1994)) but it is not likely to operate
when the alternative to EMU in 1999 is EMU one year later. Given the considerable prog
ress in terms of fiscal consolidation and disinflation that has token place especially in
the 'marginal' EMU candidates, financial markets will no longer assume that exclusion
from the first wave of EMU would mean inflationary policies for α long period. Hence it
is not likely that self-fulfilling speculative attacks of type that occurred in 1992-1 995 will
occur again. The following concentrates therefore on the criteria that are likely to consti
tute the real hurdles for the examination that will take place in 1998, namely the infla
tion criterion and the fiscal criteria.
: Β .''crrïi 's Volumel Noi 1996 17
Most economists agree that price stability and sound public finances constitute good
policy. In this sense, one should not regard the Maastricht criteria as something that has
been imposed arbitrarily. If there is α transitional cost that arises from the fulfilment of
these criteria, if should be regarded as the cost of achieving sound economic manage
ment in the long run — α desirable goal whether the country wants to participate in EMU
or not. This is also the main reason why it is of little use to discuss at length whether ful-
hlment of the Maastricht criteria is necessary from α theoretical point of view. It might
be possible to have EMU without prior convergence, but these criteria exist and they
represent sound policy.
The criterion concerning inflation deserves α bit more discussion since the transition to
price stability is sometimes painful. Most member countries are now close to price sta
bility. The overage rote of inflation in the EU has fallen continuously since 1992 and has
now gone below 3%. But there are four countries (Italy, Spain, Portugal and Greece),
whose rate of inflation exceeds the Maastricht benchmark. Further disinflation thus
remains necessary in some case. Except for Greece, these countries will need to reduce
inflation by about 2-3 percentage points. Wil l this cause more unemployment in these
countries?
Many economists assume that lowering inflation leads to higher unemployment in the
short run. It is also generally accepted, however, that the trade-off between inflation and
unemployment is not stable. It depends on the credibility of the onti-inflotionory policies,
i.e. the extent to which the reduction in inflation is anticipated and thus incorporated in
interest rotes and in wage contracts. It is also now generally accepted that α perma
nently lower inflation rote does not lead to permanently higher unemployment. On the
contrary, some recent research suggests that lower inflation is associated with better out
put performance over the long run. (See Barro, 1995, and Banian et al., 1994).
Unfortunately, it is impossible to be more precise on the size of the transitional cost of
disinflation. The so-called sacrifice ratio, i.e. the price paid in terms of higher unem
ployment in exchange for α reduction in inflation, has varied widely over the post. In
some countries, disinflation has even been accompanied from the start by more growth
and less unemployment. Existing macroeconomic models will be less reliable than in the
past, since it is likely that the behaviour of economic agents will not follow post patterns
if they see that the economic environment changes fundamentally as EMU approaches.
This is the essence of the so-called "Lucas critique" (see Lucas, 1976).
E iB E'QDC'S
18 Volume I N o Τ 1 9 9 6
It is therefore impossible to soy what transitional cost the four countries with high infla
tion would have to sustain in order to satisfy the Maastricht criterion on inflation (1 ), or whether
there would be any cost at oil. Little con thus be said about the short-term consequences
of disinflation, other than that any short-term costs ore worth the longer-term benefits.
Current forecasts indicate that the further reduction in inflation in Southern Europe should
actually be accompanied by α modest increase in growth.
Whether α country has achieved sufficient prices stability to participate in EMU will be
judged on α precise rule. The criterion on inflation is defined in the Protocol on the
Convergence criteria (Article 1 ) as:
"The criterion on price stability referred to in the first indent of Article 109j (1 ) of
this Treaty shall mean that α Member State has α price performance that is sus
tainable and on overage rote of inflation, observed over α period of one year
before the examination, that does not exceed by more than 1.5% points that of,
at most, the three best performing Member States in terms of price stability.
Inflation shall be measured by means of the consumer price index on α compar
able basis, taking into account differences in national definitions."
Contrary to the fiscal criteria there seems to be no leeway in this formulation. It would
thus be possible that α country that misses this criterion by just one tenth of α percentage
point would fail the inflation criterion. Moreover, if by chance three, possible small, mem
ber countries achieve very low inflation in 1997 the ceiling set here could be very low.
The only important qualification for the inflation criterion is the word 'sustainable'.
Presumably this means that α price performance would not be regarded as sustainable
if inflation was acceptable in 1997 but hod been higher in 1996; and was projected
to increase in 1998. However, since inflation moves rather slowly over time this is un
likely to be the cose.
The inflation criterion might turn out to be awkward since, again in contrast to the fiscal
criteria, there is very little α government con do to influence inflation in the short run.
The consumer price index reacts only sluggishly to α mix of factors that include monet
ary policy, fiscal policy, exchange rotes, wage demands, etc. There is thus very little the
authorities in the Southern European countries can do if they discover in middle to late
11 Another danger for inflation might arise for countries that enter EMU with an excessively depreciated exchange
rate. Macroeconomic models suggest that it takes a number of years before a depreciation translates into
higher prices. This raises the possibility that a country that entered EMU with stable prices but a depreciated
exchange rate would for some years have a substantially higher inflation rate than the rest of EMU. It is even
possible that such a country would no longer satisfy the inflation criterion oher it h a d already entered EMU.
However, this should not be regarded as a danger for price stability in the EMU since the countries whose cur
rencies ore relatively overvalued at the start wi l l experience a corresponding moderation of inflation.
Volumel N o i 1 9 9 6 19
The i n f l a t i o n c r i t e r i o n
m i g h t t u r n o u t t o b e
a w f c A v a r d s ince t h e r e
is v e r y l i t t l e a g o v e r n
m e n t c a n d o t o i n f l u
e n c e i n f l a t i o n i n t h e
s h o r t r u n .
1997 that their inflation is running above the Maastricht limit. By contrast, even towards
the end of that year will it be possible to take measures that hove α substantial impact
on the budget for the current year. For inflation this is not possible.
The forecast for 1997 discussed in the annex shows that Italy, Portugal and Spain will
be very close to the borderline of the inflation criterion. If inflation turns out to be better
than expected in Norihern Europe it is possible that the limit could be as low as in 1995,
namely 2.7% (1.2 plus 1.5). If at the some time growth picks up in Southern Europe and
inflation ends up only slightly higher than expected there could be α serious problem.
All in all it is thus likely that the inflation criterion will constitute serious hurdle for at least
some of the Southern European countries, but there is little governments can do at this
except watch and hope. I now turn to the adjustment in public finances which deserves
0 more extensive discussion because there ore some major decisions still to be taken.
The next section 3 turns to α careful analysis of the need for fiscal retrenchment and sec
tion 4 asks what macroeconomic consequences this will hove. The lost section turns to
the practical issues in managing variable geometry, including α concrete proposal for
the transition.
3 . T o v r a r d s s o u n d f i n a n c e s : D e b t leve ls v e r s u s c h a n g e s
It is widely assumed that the fiscal criteria imply that α country that wonts to qualify for
EMU has to hove α deficit below 3% of GDP and α public debt-to-GDP ratio of below
60%. This is not entirely correct, however, at least as far as the debt ratio is concerned.
How could this confusion arise? What ore the conditions under which α country has on
excessive dehcit? This section will try to give on answer to some of these questions.
The second paragraph of Article 104c is key in this respect:
"The Commission shall monitor the development of the budgetary situation and
of the stock of government debt in the Member States with α view to identifying
gross errors. In particular it shall examine compliance with the budgetary disci
pline on the basis of the following two criteria:
(a) whether the ratio of the planned or actual government deficit to gross domestic
product exceeds α reference value, unless
- either the ratio has declined substantially and continuously and reached α
level that comes close to the reference value;
20
; Β Popes
Volumel Noi 1996
- or, alternatively, the excess over the reference value is only exceptional and
temporary and the ratio remains close to the reference value;
(b) whether the ratio of government debt to gross domestic product exceeds α
reference value, unless the ratio is sufficiently diminishing and approaching
the reference value at α satisfactory pace.
The reference values ore specified in the Protocol on the excessive deficit pro
cedure annexed to this Treaty."
The Protocol referred to establishes the reference value for the deficit at 3% (the deficit
of general government as α proportion of GDP) and 60% for the debt (gross debt of
general government as α proportion of GDP). These are indeed the numbers that domi
nate the public discussion, but the Treaty also contains important qualifications that ore
often oveHooked.
3.1 D e b t s : Levels v e r s u s c h a n g e s
On the deficit, the Treaty could be interpreted as saying that only small overruns are
admissible and that they hove to be temporary. A valid reason for α temporary deficit
is often assumed to be α downswing in the business cycle; but there might also be un
foreseen expenditure due to α court ruling as happened recently in Italy and Germany.
It will always remain debatable what "close to the reference value" means in practice.
Is 0 deficit of 3.5, or even 4% of GDP still close? But it is now generally accepted that
the 3% deficit reference value should be considered an absolute upper limit even during
α downswing and member countries are aiming at α value somewhat lower than that
during normal times. These ore questions of detail, however, compared to the ones that
arise concerning the debt level, which in some countries is double the reference value.
In contrast to the provisions concerning the deficit, the rules concerning debt do not spe
cify that the level of debt has to stay close to the reference value. The reason for this is
quite clear: when the Treaty was negotiated, several countries already hod debt in
excess of 100% of GDP From this starting point, it was clearly impossible to get close
to the reference value in any foreseeable future because the debt level is α stock that
cannot be changed quickly. A deficit, which is α flow concept, con be adjusted rather
quickly, but it takes time for this to have on impact on the debt level. The Treaty merely
requires that the debt/GDP ratio must be moving in the right direction at α certain mini
mum speed. The decisive formulation concerning the excessive deficit issue will thus be
for the foreseeable future: "unless the ratio is sufficiently diminishing and approaching
the reference value at a satisfactory pace".
E 'Β Paoc-s
Volumel Noi 1996 21
The crucial question then becomes: What constitutes α sufficiently diminishing debt
ratio? This vogue formulation needs to be made more precise; otherwise, there will be
too much room for disagreement. The cose of Ireland, which was exempted from the
excessive deficit procedure in 1994 and 1995 — although its debt/GDP ratio was still
nearly 9 0 % — is cited in some countries os evidence that the Maastricht criteria hove
been softened. This criticism could arise only because the debt criterion is so vague.
i l l ® crwcfeJ q u e s t i o n
h&sameB: W h a t consti-
fistes β sufRcÌBTtny d i m
in ish ing d e b t r a t i o ?
It is not widely appreciated that most of the vagueness arising from the phrase "approaching
the reference value at α satisfactory pace" could actually be resolved on the basis of the
numbers contained in the Treaty, combined with some simple arithmetic (2). Box 1 shows
that 0 country that observes the 3% dehcit limit should, under ordinary circumstances
(i.e. if nominal GDP grows at 5% p.a.), see its debt-to-GDP ratio decline automatically
towards the 60% target. This is just α special cose of the general result that the debt ratio
will converge in the long run to α value that is equal to the deficit divided by the growth
rote of nominal GDP.
If the dehcit is equal to 3% of GDP, the speed of this convergence towards the target
would be slow, because only 5% of the difference between the actual debt/GDP ratio
and the 60% target would be eliminated each year. But this rule would at least ensure
α minimum of convergence, and α country that starts with α higher debt level would
automatically achieve larger reductions in the debt/GDP ratio. A country that starts with
140% of GDP would achieve α reduction of 4 % points per year, whereas α reduction
of 1.5 percentage points would result, and would be considered sufficient under this
rule, for α country that starts with α debt ratio at 90% of GDP.
2j But economists have long been aware of the simple arithmetic that follows. Kenen {1995] is just one
example.
1 1
E ·Ά P;ipc'S
Volumel N o i 1 9 9 6
Box 1. An interpretation of the Maastricht criterion on debt
The numbers specified as reference values in the Maastricht Treaty are arbitrary. The two values,
3% deficit and 60% debt-to-GDP ratio, ore at least coherent with each other, however, if one
assumes that nominal GDP grows at 5% per year. This seems α reasonable assumption since it
corresponds to the growth rate that α relatively good performer in terms of price stability, such as
Germany, experienced during the 1980s. (During the 1960s and 1970s, nominal GDP actually
grew at over 8% in Germany.) If growth in the EU stays at 3% (i.e. just α bit above potential out
put growth), α 5% nominal growth rate would be compatible with inflation of 2% (less than the
German average over the last 40 years).
Given this assumption, the two reference values are consistent with each other in the sense that
at α 60% debt/GDP ratio and α 3% deficit will leave the debt ratio unchanged. This can be seen
by considering the government budget constraint in terms of ratios of GDP, which implies that the
change in the debt ratio, denoted by b, - b,.,, is approximately equal to the deficit (the overall
deficit, not the primary deficit), indicated by def,, minus an adjustment factor for GDP growth:
(1) b, - b|..| = def, - b, * growth of nominal GDP
If the nominal GDP growth is 5%, this equation implies that the 3% deficit limit will lead to α debt-
to-GDP ratio automatically at 60% since if the deficit, def,, equals 0.03, equation (1) can be
rewritten as:
(2) b,-b,.| = - 0 . 0 5 * jb,-0.6)
If the debt ratio is initially above 60%, it will decline, and vice verso if it starts out below 60%. It
will be constant only if b, = 0.6 (i.e. 60%).
This result depends, of course, on the assumption of α residual inflation rote of about 2% (plus
real growth of 3%). With absolute price stability, GDP would grow only at 3%; in this case, α deficit
of only 1.8% of GDP would be required to keep fhe debt ratio constant. If one takes into account
the average deficit should be below 3% of GDP since this value is the upper limit, there is thus
room for improvement on the debt ratio even if nominal GDP grows by less than 5%. It the budget
were balanced in good times and allowed to go to α deficit of 3% in bad times, e.g., the overage
deficit would be 1.5% of GDP if good times occur with the some frequency as bad ones.
The most important implication of equation (2) in this context, however, is that one-twentieth (0.05)
of the discrepancy between the actual debt ratio and the Maastricht target would be automati
cally eliminated each year, if the deficit is 3% of GDP
This suggests that the expression in Article 104c, 2b that α debt/GDP ratio above 60% consti
tutes an excessive deficit "unless the ratio is sufficiently diminishing and approaching the reference
value at α satisfactory pace" could be interpreted more precisely as saying that the debt ratio
should be declining at least by enough to reduce the distance between the 60% reference value
and the starting point by at least by 5% p.a. If this rule is accepted, any government whose deficit
was below 3% of GDP (and that does not accumulate debt off-budget) would automatically meet
the debt criterion.
t IB Pnpi^b
Volumel Noi 1996 23
In order to ensure that the improvement is not transitory it would be necessary to apply
this criterion over α number of years. This could be achieved, for example, through the
following practical rule that could be adopted informally by ECOFIN:
The debt to GDP ratio is considered "approaching the reference value at α satis
factory pace" if, over the last three years, it has been declining continuously, and
the total reduction has been equal to three-twentieths of the difference between the
debt ratio at the beginning of the three-year period and the reference value of 60%.
This rule should be applied each year when the public finances of member countries are
examined for the excessive deficit procedure. In practice, it would be relevant for those
member countries that hove α debt/GDP ratio that is clearly above the 60% reference
value. It should be obvious, but it bears repeating, that this rule should also apply after
α country has joined in order to ensure continued movement towards the 60% target.
The main reason why even such α slow (at least at hrst sight) speed of adjustment should
be acceptable is that the potential pressure on the ECB that derives from α large debt
level is much reduced once financial markets see that the debt/GDP is clearly on α
durable downwards path.
The rule proposed here is based on the assumption of α growth rote of nominal GDP of
5% p.a. But the ECB might do better than the Bundesbank and keep nominal GDP growth
fo below that figure. This does not mean that one should change the rule for judging
satisfactory progress of the debt ratio towards the reference value of 60% of GDP. If
nominal GDP growth dropped to 3%, α deficit of 3% of GDP would lead the debt ratio
to converge to 100% of GDP This is clearly incompatible with the Treaty. The reduction
in the debt ratio proposed here should thus be understood as the minimum that has to
be achieved whatever the growth rote of nominal GDP
Table 2 below shows the evolution of the debt/GDP ratio until 1996, for those 8 mem
ber countries that ore clearly above 60%. It is apparent that, except for Ireland, there
has been no improvement since 1993. The lost column of this table then shows what
would be required if the above-mentioned rule were applied in the 1998 examination
that will be based on budgetary data for 1997 (assuming the three year time-span
1994-1997 is the basis). Belgium would be admitted if the ratio hod declined from only
about 135 to 123% of GDP; this would require α considerable effort, however, since
the ratio has essentially been constant over the last few years. For Ireland, the speed of
decline since 1993 (12 percentage points of GDP until 1995) would be sufficient to
continue to be exempt from the excessive deficit procedure (3). For countries with α debt
ratio of around 70%, the required adjustment would be minor.
3) The picture for Ireland would change if one were to take 1992 as a storting point. The development in 1993,
however, was due to special circumstances.
f IB Papers
2 4 Volumel N o i 1 9 9 6
It is noteworihy that neither the Maastricht Treaty nor any other official document uses
the accounting identity that the increase in government debt over any given year should
be equal to the dehcit incurred during that year so that the dehcit and the debt criteria
ore linked. Unfortunately, this is seldom the case in reality. Small deviations from this
accounting equality would not matter. But the discrepancies that ore contained in the
official figures for the recent post ore so large that they moke the interpretation of the
convergence criteria very complicated.
For example, in 1994, the dehcit of Germany was below 3% of GDP so that the debt/GDP
ratio should have been approximately constant given that the growth of nominal GDP was
slightly above 5%. But the debt ratio increased during 1994 by about 8.5% of GDP (i.e.
more than one would hove excepted on the basis of the evolution of nominal GDP and the
dehcit). The reason was that the German government took over the accumulated liabilities
of the Treuhandonstolt, the agency charged with privatisation in the former GDR.
Table 2. Public debt in high-debt countries (Percentage of GDP)
Public debt as % of GDP 1 9 9 3 1 9 9 4 1 9 9 5 Required*
in 1 9 9 7
The d e b t t o GDP r a t i o is
c o n s i d e r e d " a p p r o a c h
i n g the r e f e r e n c e v a l u e
a t a s a t i s f a c t o r y p a c e "
i f , o v e r t h e l a s t t h r e e
y e a r s , i t h a s b e e n
d e c l i n i n g c o n t i n u o u s l y ,
a n d t h e t o t a l r e d u c t i o n
h a s b e e n e q u a l t o
t h r e e - t w e n t i e t h s o f t h e
d i f f e r e n c e b e t v / e e n
t h e d e b t r a t i o a t t h e
b e g i n n i n g o f t h e t h r e e -
y e a r p e r i o d a n d t h e
reference v a l u e o f 0 0 % .
Belgium
Denmark
G r e e c e
Ireland
Italy
Netherlands
Portugal
Sweden
37.5
80
115
97
119
81
67
76
135
76
113
91
125
78
69
80
134
74
114
86
125
78
71
81
123
74
105
86
115
75
68
76
Source: European Commission a n d o w n calculations,
* To achieve α reduction of 3 / 2 0 with respect to 1 9 9 4 .
The prize for the largest "stock flow adjustment" in recent times goes to Greece, where
it exceeded 20 percentage points of GDP in one year alone (1993)1 In coses like this
one, clearly the deficit numbers ore not sufficiently informative by themselves and the for
mulation in Article 104c(2) that "the Commission shall monitor the budgetary situation
and the stock of government debt in member countries with α view to identifying gross
errors" acquires real meaning. Reconciling dehcit and debt hgures should be worth α
major effort by the services of the Commission (4).
4 j It is surprising that there has been no official explanation of these inconsistencies and no official comment
on them in terms of the interpretation of the fiscal convergence criteria. They must clearly be taken into account
during the excessive deficit procedure. An increase of the debt ratio could in some cases be cause for concern,
but in other cases they might be accepted as having little to do with the overall fiscal situation as explained in
Gros ( I 9 9 o a j .
•'B Ρ α ι : ο · ΐ
Volumel N o i 1 9 9 6 2 5
Under EMU, most of the legitimate reasons for the stock flow adjustment (e.g. borrowing
by the central bonk to bolster its reserves, α change in the domestic value of debt de
nominated in foreign currency, as long as it is in another EU currency, due to α devalu
ation) should disappear. The practice of keeping certain items off-budget should then be
thoroughly scrutinised.
W h a t e v e r r u l e is f i n a l l y
r e t a i n e d , i t is i m p o r t
a n t t h a t t h e i n t e r p r e
t a t i o n o f t h e c r i t e r i o n
b e c l a r i f i e d i n a d v a n c e .
All in all, these considerations imply that α government that keeps fhe deficit at, or even
slightly below, 3% should also be able to satisfy the debt criterion, provided that it does
not accumulate debt off-budget and that growth is satisfactory. The dehcit is thus the key
variable even for countries whose debt level is, still, for above the target value of 60%.
This might also be ultimately the reason why the Treaty speaks only of an excessive dehcit
procedure.
What would be the implications of the proposed rule in reality? A dehcit of 3% of GDP
at present is not enough to obtain the required reduction in the debt ratio for some mem
ber countries for two reasons: i) the growth rote of nominal GDP is slightly below 5%,
and, more importantly, and ii) off-budget debt accumulation is continuing to the tune of
1 to 2% of GDP each year. This implies that countries with α debt level for above 60%,
such OS Belgium and Sweden, would need to aim at α dehcit (os ofhciolly measured)
substantially below 3% of GDP. For these countries, dehcits not exceeding 1-2% of GDP
would be the only way to attain the minimum reduction in the debt level outlined above.
Box 2 shows how the debt/GDP ratio would behave if the automatic stabilisers are allowed
to work so that dehcits ore below 3% of GDP during good times.
The proposed rule is only one possibility. An alternative solution would be to require fhe
primary dehcit (i.e. the dehcit net of interest charges on public debt) to be consistent with
α decline of the debt ratio under normal real growth and real interest rate conditions.
This would have the advantage of making the assessment of the hscal situation inde
pendent from 0 variable that the government does not control, namely the long-term
interest rate, but at the cost of making the evolution of the debt ratio less predictable.
Any rule based on the primary deficit would also be difficult to reconcile with the Treaty's
emphasis on the overall dehcit.
Whatever rule is hnolly retained, it is important that the interpretation of fhe criterion be
clarihed in advance, because this would: i) increase the credibility of EMU, ii) ovoid α
situation in which some countries ore (or appear to be) treated more leniently than others
for political reasons; and iii) help countries to design their convergence programme,
including those that will not qualify in 1998, but aim to do so as soon as possible there
after.
26
! 3 Ραο;?·5
Volumel N o i 1 9 9 6
Box 2. The business cycle and automatic stabilisers
130
120
110
100
Ü 90 I ο 80
SO I I I I I I I I I I I I I I 1 I I I I I I I I I
1 2 3 4 5 6 7 8 9 10 Π 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Years
^ ^ ^ — With constant 3% deficit and no business cycle (constant growth of nominal GDP).
Wifh automatic stabilisers [3% deficit during downswing, 1 % during upswing).
The pract ical rule proposed here to determine whether α d e b t / G D P ratio is " a p p r o a c h i n g the ref
erence value at α satisfactory speed" should be considered as the absolute minimum. If α country
w e r e to have α deficit of exactly 3 % of GDP, g r o w t h of nominal GDP of 5 % a n d no accumulat ion
of debt off-budget the movement of the d e b t / G D P ratio towards the 6 0 % reference value w o u l d
indeed be slow as shown in the a c c o m p a n y i n g g r a p h . Even after 3 0 years the rat io w o u l d still
be a b o v e 7 0 % .
However, there is now b r o o d agreement that the deficit should in general be held clearly below
3 % in order to ensure that even in α recession it does not exceed this limit. It is thus more realistic
to assume that the deficit wi l l sometimes be below 3%. The lower line in the g r a p h shows what
w o u l d happen if the deficit were to be equal to 3 % of GDP dur ing α d o w n s w i n g of the business
cycle (defined here as α growth rale of nominal GDP of 4%) a n d 1 % dur ing the upswing of the
business cycle (defined here as α g r o w t h rote of nominal GDP of 6%). This still implies an average
deficit over the cycle of 2 % of GDP, but the automatic stabilisers w o u l d be a l l o w e d to work. Under
these condit ions, the d e b t / G D P ratio w o u l d g o d o w n in α stepwise pattern because dur ing the
d o w n s w i n g two effect reinforce each other: the deficit is large a n d the denominator (GDP) g r o w s
more slowly. Under this a p p r o a c h , the movement towards Ihe reference value is much quicker.
Starting from α d e b t / G D P ratio of 1 2 0 % , it w o u l d take "only" about 1 5 years to get close to 7 0 % .
EIB Papers
Volumel N o i 1 9 9 6 2 7
For a n u m b e r o f c o u n
t r i e s a d e f i c i t o f 3 % o f
G D P is n o t s u f f i c i e n t t o
p u t t h e d e b t l e v e l o n a
c l e a r l y d o v ^ n w a r d
p a t h , b e c a u s e n o m i n a l
GDP g r o v ^ h is l o v / e r
t h a n 5 % a n d t h e r e is
s o m e o f f - b u d g e t d e b t
a c c u m u l a t i o n .
3 . 2 . Def ic i ts: H o w m u c h a d j u s t m e n t is n e e d e d ?
Satisfying the dehcit criterion alone already implies α considerable hscal adjustment in
0 number of member countries. Table 3 below shows the most recent estimates of gen
eral government dehcits in 1996 and forecosts for 1997. The hgures for 1990 ore also
presented to show that the situation in 1995 will be similar to the situation just before
Maastricht was negotiated. It is remarkable that in one key country, France, on unpre
cedented deterioration in the hscal accounts took place between 1990 and 1995, i.e.
just offer the country signed up fo observe fhe Maastricht fiscal convergence criteria. The
largest swing took place in Sweden, where the deterioration between 1994 and 1990
was equivalent to almost 15% of GDP but the improvement planned for 1996 is also
very large. Finland experienced similarly large swings between 1990 and 1995.
Table 3 below shows that by late 1996 all member states except the UK, Italy and
Greece were expected to reach α dehcit of 3% of GDP by 1997. If these plans are carried
out this would lead to on overage dehcit for the EU 15 of 3% for that year. This is α
remarkable improvement over 1995, when the overage was equal to 5% of GDP
The forecast for 1997 are mode under the assumption of unchanged policies. They thus
show what would happen if no further action is token and the economy expands mod
erately (by 2.3% in 1997 after 1.8% in 1 996). But even if one somewhat discounts the
plans for 1997, which must always be treated with caution, it is clear that the need for
hscal adjustment in most member countries is now small - if one assumes that α deficit
of 3% of GDP is the target.
One has to keep in mind that for α number of countries α dehcit of 3% of GDP is not
sufficient to put the debt level on α clearly downward path, because nominal GDP growth
is lower than 5% and there is some off-budget debt accumulation. This might be the case
for Italy where the debt ratio will decline only from 1 25 to 1 23% even if the plans ore
carried through because some of the measures to reduce the dehcit recorded in 1997
will not affect public debt accumulation. And the Netherlands will actually experience
on increase in its debt ratio unless α further effort is undertaken as indicated in the lost
column of this table. The coses where the need for further efforts to obtain α 'sufficient'
reduction of the debt ratio is more doubtful ore indicated by α question mark.
Belgium might use some privatization receipts to reduce its debt ratio in 1997, but this
should not be counted here since what matters is fhe capacity fo reduce fhe ratio con
tinuously over the long run. For Belgium, α further hscal adjustment of about 1.5 percen
tage points of GDP (to reach α dehcit of about 1.5% of GDP) and for the Netherlands,
α further effort of about 1 percentage point (to α dehcit of about 2% of GDP) would prob-
28
E IB Poptys
Volumel Noi 1996
ably be required to achieve α sufhcient reduction in the debt levels in terms of the prac
tical rule discussed above.
Table 3. Public sector dehcits in the EU
The C o m m i s s i o n c a l c u
l a t i o n s m i g h t const i tute
a n o v e r s t a t e m e n t o f
t h e n e e d f o r f i s c a l
r e t r e n c h m e n t , s ince
t h e s e c o u n t r i e s a l s o
p a y m u c h h i g h e r
i n t e r e s t r a t e s . A s t h e y
g e t c l o s e r t o E M U , t h e
d e b t s e r v i c e b u r d e n
s h o u l d f a l l c o n s i d e r
a b l y .
Denmark*
Ireland*
Luxembourg*
Finland
Netherlands
Austria
Belgium
France
Germany
Greece
Italy
Portugal
Spain
Sweden
UK
EU 15
1990
1.5
2.2
-5.9
-5.4
5.1
2.1
5.4
1.6
2.1
14.0
10.9
5.5
3.9
-4.2
1.5
Deficit as % of GDP Maastricht Defir
1995 (final)
1.6
2.0
-1.5
5.2
4.0
5.3
4.1
4.8
3.5
9.1
7.1
5.1
6.6
8.1
5.8
5.0
ition
1996 (Provisional)
1.4
1.6
-0.9
3.3
2.6
4.3
3.3
4.0
4.0
7.9
6.6
4.0
4.4
3.9
4.6
4.4
1997 (Planned)
0.3
0.9
-0.5
2.2
2.5
3.0
2.9
3.0
2.9
6.5
3.3
2.9
3.0
2.9
3.5
3.0
Fürth sr adjustment needed in 1997 to reach sufficient progress towards 60% debt ratio
none
none
none
none
1.0
?
1.5
none
none
3.5
1.5
?
?
?
none
0.8
Source: Commission of the European Community, * country is no longer in the excessive deficit procedure as of 1996.
There are, however, some countries that would hove to reduce their deficits by much
more than 3 percentage points of GDP Italy and Sweden stand out with an adjustment
need (os of 1 995) of about 4% of GDP to reach α deficit of 3% of GDR However, fhe
figures used so for might constitute an overstatement of the real need for hscal retrench
ment, since these countries also pay at present much higher interest rotes on their public
debt. As they get closer fo EMU, the debt service burden should fall considerably. Would
this effect be important? The answer is yes, but unfortunately it is difficult to predict the
exact amount of the interest rate savings. Moreover, these savings will be available only
wifh α certain time log as old debt is retired. Interest expenditure for 1997 was already
determined to 90% at the end of 1996 through the medium and long term debt issued
in the post (i.e. mainly 1 995 end 1 996). In countries with α shorter average duration of
; !B Pa[:i."s
Volumel Noi 1996 29
public debt (shori term debt or floating rate debt being important) the impact in 1997
would be higher.
Table 3 below uses an overage of the estimates of the interest savings the highly indebted
countries could experience as they converge to EMU (from Gros (1996α) in order to assess
the reduction in non-interest expenditure that remains necessary to reach the 3% limit.
Table 4. Fiscal adjustment and debt service savings
Basic Data (1995)
Greece*
Italy*
Portugal
Spain
Sweden
Debt/ GDP
114
125
71
65
81
Deficit/ GDP
9.3
7.4
5.4
6.0
7.0
Potential debt service savings
6.7
4.5
. 5 - 3 . 0
1.5
2.2
Adjustment needed (in % of GDP) to reach α 3% deficit
Overall deficit
7.3
5.4
2.4
3.0
5.0
Primary balance with debt service savings
0.6
1.0
2.0-0.0
1.5
2.8
* For these countries, the target is α deficit of 2% of GDP because of the high debt level.
Source: Own calculations; see Annex 2 in Gros (1996α).
This table suggests that, if they could count on EMU interest rates, even countries like Italy,
Spain and Portugal would have about the some need for discretionary adjustment in pri
mary (non-interest) expenditure (and/or foxes) as other countries with lower dehcits, i.e.
about 1 - 2% of GDP The remainder would come through lower debt service costs. For
Sweden, the need for adjustment, after taking into account the full potential for α reduc
tion in the cost of servicing public debt, is now only moderately above that of the rest.
A closer analysis of the hscal accounts thus suggests that the high-interest rote member
countries could be in α sort of "catch-22". As long as they ore outside EMU, they need
to cut dehcits by between 3 and 5% of GDP But if they were in EMU, the need for hs
cal adjustment would be manageable, about 2% of GDP This central problem will
appear again in the lost section.
3.3. Deficits: the out look for 1997
As with inflation, it will be useful to discuss briefly the outlook for 1997 since the evolu
tion hscal dehcits will be continuously monitored by hnanciol markets which will also
30 E :B Pric^.-s
Volumel Noi 1996
Past f o r e c a s t s w e r e
r a t h e r a c c u r a t e , a n d
o n e s h o u l d t a k e t h e m
s e r i o u s l y
adjust their expectations in function of the new forecasts that will become available from
time to time. The examination that will determine the list of first wove EMU countries will
take place in early 1 998 and will be based on the hgures available then. It will thus be
useful to compare the forecasts that were made in the past to the outcomes. The annex
does this for hve member countries: France and Germany (which ore key for starting
EMU) and the three southern European countries which ore likely fo be 'borderline' can
didates in 1998.
The very brief analysis of the annex suggests that in the post forecast were rather accu
rate and that one should thus take fhe forecasts for 1997 presented in Table 3 seriously.
These forecasts ore usually mode under the assumption of unchanged policies and use
rules of thumb from the post to link expenditures and revenues to economic activity. This
is not justihed for the year in which oil governments will use all means to show α dehcit
under fhe Maastricht dehnition of below 3% of GDP These forecasts must thus be read
OS indicators of the need for additional measures, not os predictions of actual outcomes.
By contrast, there is little governments con do to influence inflation in the short run. The
private sector should adjust its behaviour only moderately in the expectations of EMU
when the economy has already settled down to α low inflation equilibrium (5).
For hscal deficits one has thus to use Goodhord's law which one could adapt to this cose
to say that an economic indicator that becomes politically as important os the dehcit will
behave completely differently than in the past.
4 . The m a c r o e c o n o m i c s o f f isca l r e t r e n c h m e n t : W i l l f u l f i l m e n t o f t h e
M a a s t r i c h t c r i t e r i a h u r t e c o n o m i c a c t i v i t y ?
What would be the short-term macroeconomic consequences of implementing the deh
cit reduction identihed above as necessary to observe the hscal criteria? In most macro-
economic models (see Hughes Holleff and Peter McAdam, 1995, for further references),
α fiscal contraction leads to lower demand and hence to lower output growth and
employment creation, of least in the short run. These effects con in principle be quonfihed
with some precision. Most of the existing macroeconomic models hove multipliers
somewhat above one so that they would predict that if any one country reduces its deh
cit by 2 to 3% of GDP demand would fall by α at least similar amount in the short run.
If several countries move hscal policy in the some direction at the same time, the impact
on output might, however, be different depending on the sign of the spillover effect.
Standard macroeconomic models generally predict that α fiscal adjustment quickly
causes α fall in output that is then followed by α rebound. This "U-form" adjustment pot-
5 j In other words the scxalled 'Lucas critique' should in reality not invalidate forecosts biased on past behavioural
relationships.
Volumel N o i 1 9 9 6 3 1
tern appears in most models. However, the different large macroeconomic models that
are available differ strongly in their implications for the length and the strength of the in
itial dip. There ore also large differences concerning the subsequent recovery, that,
depending on the model, could lift output above the stariing point because of α strong
interest rate effect. The short-term macroeconomic consequences of α fiscal adjustment
ore therefore difficult to determine with any precision.
The simulations from the early 1990s that ore available ore difficult to compare with
more recent exercises because the adjustment required (to reach α dehcit of 3% of GDP)
has changed over time. It is symptomatic that none of the international organisations
(neither the IMF nor the OECD) nor the services of the Commission has so far published
the results that one would obtain with their macroeconomic models. At any rote, differ
ent simulations also make different assumptions concerning the time path of the adjust
ment, the accompanying monetary policies and the number of countries involved. For
example, one simulation might hold exchange rotes and interest rotes constant whereas
another allows for flexible exchange rotes even within Europe. Box 3 discusses the result
of some recent simulation exercises performed for the CEPS Economic Policy Group.
Hughes Hallett and McAdom (1995) provide the most recent evaluation of the macro-
economic effects of the hscal adjustment necessary to achieve α deficit of 3% of GDP in
some larger member countries (France, Italy, and the UK). They use the model of the IMF
to determine the impact of this adjustment to the Maastricht norm on output and prices.
Their main result is that the models predict α fall in output and prices (compared to the
so<alled baseline) that could be substantial and protracted, depending on the timing of
the adjustment. The cumulative percentage decline in output would in general be
somewhat smaller than the reduction in the dehcit as α percentage of GDP. Thus for
France, the fall in demand could be somewhat below 2%, whereas it would be larger in
Italy or the UK.
t ITJ t"cifx:;b
32 Volumel N o i 1996
Box 3. The effects of large fiscal retrenchments: What do macroeconomic models tell us?
Simulations with standard macroeconomic models that assess the adjustments necessary to meet
the 3% deficit limit suggest that the remaining reduction in deficits (about 2% of GDP in France,
somewhat more in the UK and, about 4% of GDP in Italy) involve non-trivial costs in terms of out
put losses. The results differ depending on the model used and the starting point, but even more
optimistic simulations conclude that the fiscal multiplier (measuring the decline in output per unit
of deficit reduction) typically lies between 0.5 and 1.0 for a large country. This implies that fiscal
retrenchment will significantly dampen growth in several European countries in the run-up to EMU.
Eichengreen and von Hagen (1995), for example, report serious recessions resulting from α
contraction of government spending, where output recovery takes up to five years. One reason
for the slow recovery is that the deficit criterion is formulated in relation to GDP. This implies that
expenditure needs to be cut by more than needed initially to reach the 3% target (ex-post) to offset
the fall in GDP growth. This aggravates the recession, making the needs for retrenchment even
larger.
One reason why the models usually produce large output costs is that they assume that nothing
else affects the monotonie relationship between fiscal policy and output. This is often not the case,
however. Pisani-Ferry and Cour (1995) study large budgetary adjustments in OECD countries
since the 1970s and report that the output costs in their sample have been rather low. They there
fore conclude that the output costs con be lower than is generally assumed.
One possible explanation for this finding of low impact of fiscal retrenchment could be that the
risk premium on the public debt diminishes with α sustained improvement of the deficit. With the
exception of Denmark, however, the impact of the adjustment on interest rates has in reality been
rather low in the short run. One thus has to attribute the low negative effects of α determined fis
cal retrenchment to other stimulating monetary policy and/or to the direct expansionary effect the
restoration of sustainability have on private demand.
Hughes Hallett and McAdam (1995) used the IMF's MULTIMOD model to study the effects differ
ent adjustment scenarios would hove on the long-run growth in those countries undertaking them.
They point out that it might be quite easy for countries to squeeze in by forcing their deficit down
to 3% but that it would be harder to keep it there. The standard models also predict α consider
able deflation (a fall in the price level relative to the baseline) following fiscal retrenchment. Since
this is not needed in most countries (especially France), these models thus also imply that α re
strictive fiscal policy could be coupled with α less restrictive monetary policy without endangering
price stability. By keeping the price level at the baseline, α large part of the output loss could be
avoided.
E !B Popes
Volumel Noi 1996 33
Pisani-Ferry (1996) and Giovozzi and Pagano (1995) also suggest that the experience
with large hscal adjustments in the post shows that determined action con reduce, and
possibly even eliminate the output cost of hscal contractions. Indeed, most of the sustained
hscal adjustments that occurred in OECD countries over the lost two decades caused
actually very little or no output losses at oil. The explanation might be that α hscal adjust
ment that is the result of α change in the fundamental rules followed by the government
should hove different implications than on action that just changes the dehcit for α
couple of years, as argued in Lucas (1976). EMU would certainly représenta shift in fhe
fundamental rules for both hscal and monetary policy.
Should one therefore accept fhe conclusion that fiscal retrenchment leads to α temporary
reduction in economic activity? The basic problem with the macroeconomic simulations
is that 0 fiscal adjustment to satisfy the Maastricht criteria should have other than the
standard demand effects. In countries where the present fiscal situation is not sus
tainable, the interest rote refiects also the fear of hnanciol markets that the government
will in the end be forced to renege on its debt or create some surprise inflation. This is
presumably the reason why, with one exception, namely Belgium, the countries with the
largest adjustment need ore also those that hove the highest risk premium in the sense
that fhe interest rote differential with Germany is high. For these countries (mainly Italy
and Sweden), it is virtually certain that α decisive hscal adjustment will lead to much
lower interest rates. It is difficult to soy how large this conhdence effect will be, but the
potential is certainly considerable if one takes into account that fhe interest rote differ
ential between Italy and Germany has hovered around 5 percentage points in 1995.
Moreover, as shown above, once interest rotes fall, the need for adjustment on the pri
mary budget is much reduced. Hence, α confidence effect on interest rates would be
benehciol on two accounts: it boosts demand and reduces the need for fox increases
ond/or expenditure cuts. These considerations suggest that one should distinguish be
tween those countries where the dehcit is above 3% of GDP but the debt level is low,
and those countries where the situation is unsustainable because the debt-to-GDP ratio
is already high and would increase even further without corrective action.
For countries like France, Holland, Belgium and Austria, the interest rote differential with
respect to Germany is small. (For most of 1995, it was around 70-80 basis points for
France on short-to-medium-run maturities.) This suggests that for these countries, the confi
dence effect through lower interest rotes might be weak. France is the most important
example for this situation. In this case, the consequences predicted by macroeconomic
models might be qualitatively acceptable.
By contrast, the economies of the high-debt countries (notably Italy and Sweden) ore likely
to react differently. Decisive action to put the debt/GDP ratio on α clearly declining path
FIB Ptjpors
3 4 Volumel N o i 1 9 9 6
should lead to α large foil in interest rotes. Although the need for fiscal adjustment might
of first sight be larger in Italy and Sweden, the implementation of decisive measures
might hove less of α negative impact on output and demand than the more modest
adjustment required in the low-debt countries.
It is not impossible that α convincing fiscal adjustment could actually exert on exponsion-ory
effect on demand and output in countries that start from α truly unsustainable situation. A
decisive reduction in the deficit that stabilises public debt or even puts it on α downwards
path could lead to so much lower interest rotes that their positive effect on demand might
more than offset the direct impact of higher foxes or lower expenditure (6). This happened
in the 1980s in Denmark and Ireland where α sharp fiscal contraction was accompanied
by on expansion of output. The importance of the confidence effect con also be seen (op
erating in the opposite direction) in the experience of Spain and Italy discussed above.
Domestic demand, including investment, fell sharply in both countries after the 1992 ERM
crisis, despite small reductions in interest rates. This suggests that α confidence effect con
operate even in the absence of any interest rote changes. Giovozzi and Pagano (1995) and
Sutherland (1995) provide empirical support and analytical models for this point of view.
Even if one were to accept the predictions of the macroeconomic models that hscal
retrenchment leads to α recession or α protracted slowdown of growth, it would be one
sided to emphasise the macroeconomic costs of the Maastricht criteria. These criteria
simply represent generally accepted principles of sound hnonce. Maastricht con thus be
held, at most, responsible for encouraging governments to moke hard choices that they
would eventually hove hod to moke in any case.
fn general, the gains
f rom EMU increase
wi'fft fhe number of
participants.
5. H o w to manage var iable geometry?
There is little point in discussing whether variable geometry is desirable or not. The port
of the Maastricht Treaty dealing with EMU is clearly based on the assumption that the
third, decisive stage, of EMU may initially involve only α subset of countries. This is implicit
in the provisions that stipulate that EMU should start by 1999, but that only the countries
satisfying the convergence criteria con participate. Those that do not satisfy the con
vergence criteria will hove α derogation. In practical terms, α derogation means that the
country concerned does not link its currency irrevocably to the others and does not par
ticipate in the formulation and conduct of the common monetary policy.
The position of the countries with απ opt-out clause, i.e. the UK and Denmark, would for
mally be similar to that of the ones with α derogation. Politically, and economically, their
positions would, however, be quite different. Expressed in colloquial terms, it is the dif
ference between the "willing, but temporarily unable" and the "unwilling". We will first
6j Recent theoretical contrihiutions (see Giavazzi a n d Pagano, 1995} have also shown that if the fiscal situa
tion is unsustainable, a fiscal adjustment can be expansionary.
I '3 Pnr^e's
Volumel N o i 1 9 9 6 3 5
discuss the general problems raised by variable geometry and then present α concrete
proposal for the "willing but temporarily unable".
5 . 1 . Variable geometry in general
What ore the institutional, economic and political consequences of the form of variable
geometry foreseen in the Treaty? Does it represent α serious hurdle for EMU, especially
for those "willing, but temporarily unable"?
The institutional problems of variable geometry are already addressed in the Treaty
which regulates explicitly how the ESCB (and to some extent ECOFIN) will hove to deal
with α variable number of participants in the third stage. The initial group of countries
will, of course, have more influence than the latecomers on the institutions since the
Treaty stipulates that only the countries that participate in the initial group will decide on
the composition of the Executive Board (the President, Vice-President and four other mem
bers) of the ECB (7). Unless it is decided that not all 6 members ore appointed immedi
ately (this possibility is already foreseen in the Treaty), these 6 members of the Executive
Board will initially hove α very important position in the Council of the ECB because
there may be only 6 to 8 presidents of the notional central bonks participating from the start.
The initial group of countries participating in EMU will also take certain decisions
concerning the implementation of the common monetary policy and the instruments to
be used by the ECB that will hove to be accepted by the countries that join later. These
ore minor issues, however, compared to the economic and political ones.
It is tempting to compare the transitional arrangements for EMU to those that hove
always been mode for weaker member countries (and new members). There is one cru
cial difference, however. In the legal held, the transitional periods con be dehned and
limited exclusively in terms of time. In the monetary held, this is not possible, because
what counts ore the results, i.e. the fulhlment of the convergence criteria — and this cannot
be guaranteed by the passage of time.
But what ore the incentives created by the Maastricht provisions? In general, the gains
from EMU increase with the number of participants. The old adage "the more the mer
rier" applies thus to EMU as well. However, once α core group that includes at least
France, Germany and some other countries has started EMU, the marginal gains in strictly
economic terms from adding "peripheral" countries are small, because the latter account
only for α small shore of trade and output of the Union. This is not necessarily true the
other way round: for α "peripheral" country, the gain from participating in EMU might
be very large indeed. In general, this implies that the nature of EMU will be affected by
7 J S e e A r f c f e /09l<5.
E;IB Papers
3 6 Volumel N o i 1 9 9 6
the composition of the group of countries that takes the hrst step. The latecomers will
hove to adapt to what has already been decided for them (see Pisani-Ferry (1996)).
The asymmetry in interests between the core and the rest could create problems. For
example, it has been argued (see De Grauwe, 1995) that if the inifiol core group per
ceives that it has α higher preference for price stability than the rest of the EU, it might
be tempted to increase the requirements for subsequent participation by the countries
that ore perceived to be weaker. In this view, it would even be possible that the coun
tries that cannot participate in EMU from the start might be excluded indehnitely. It is dif-
hcult to see how this could come about in reality, however, since the convergence criteria
ore dehned in objective terms. For the only exception, i.e. the rule on public debt, we
hove shown above that it is possible to devise α rule of thumb that would eliminate most
arbitrariness from this criterion as well.
A more important reason why variable geometry is likely to create problems is that the
countries that ore not port of the initial core might fear more hnanciol market pressures
if EMU goes ahead without them. The reaction of hnanciol markets to the remarks by the
German Minister of Finance in October of 1995 shows that news about the likelihood
of participation in EMU can hove important consequences. The most obvious impact was
on Italy whose currency depreciated immediately by several percentage points and inter
est rotes rose by more than one percentage point.
But there were also some "dogs that did not bark" in the autumn of 1995 (and other crisis
episodes) which ore equally important. Firstly, although Mr. Waigel also mentioned
Belgium as one of the countries that was unlikely to join EMU in 1999, there was no
reaction in the Belgian financial markets. The reason must be that the markets expected
that the hard currency policy of Belgium will be continued whether or not the country is
part of the core founding group of EMU. Apparently hnanciol markets suspected in
1995 that this might not be the cose for Italy in the sense that the efforts to stabilise
public hnonces would weaken if Italy's participation in EMU had to be postponed. An
immediate practical consequence was also that the re-entry of the lira into the ERM,
which seemed imminent, was postponed sine die. Secondly, there was also no reaction
in the hnanciol markets of those countries that hove clearly no prospect (or intention) of
joining EMU in 1 999. For these countries, there was no new element that could hove
induced hnanciol markets to reassess their evaluation. It is interesting to contrast this epi
sode to the non-reoction of hnancial markets to remarks by the President of the
Bundesbank who recently made similar comments. They were not taken into account by
hnanciol markets because the perception is now that even if Italy did not qualify in 1998
it would pursue its adjustments and qualify shortly thereafter.
E IB Papers
Volumel N o i 1996 37
Countries that are
close to fulf i l l ing the
criteria, but are not
admit ted into the first
g r o u p because they
fa l l a l ittle short can
expect a strong reac
tion f r o m their f inan
cial markets.
The 1 995 episode thus suggests that the uncertainty about the composition of the core
group is likely to lead to increased financial market volatility for those countries that
would like to participate in EMU and undertake adjustment efforts mainly for that pur
pose. If markets perceive that the efforts ore undertaken not because fiscal adjustment is
desirable in its own right, they might conclude that if the interest rote is high the country
will not be able fo participate in EMU (because fhe deficit will be high). This attitude might
also lead to α slackening of the adjustment efforts. However, another equilibrium is also
possible: if interest rotes ore low, the adjustment required to enter EMU is smaller and hence
it is more likely that it will actually be undertaken. Italy in 1995 seems to exemplify the
"bod" equilibrium, whereas Belgium seems to exemplify the "good" equilibrium (8).
All this implies thot for some countries, hnanciol market uncertainty will persist until the
European Council takes its hnol decision scheduled for 1998. Until then, hnanciol mar
kets could oscillate between the two equilibria causing large swings in exchange rotes
and interest rotes.
The problem will be most acute when the decision has to be taken in 1998. Countries
that ore close to fulhlling the criteria, but ore not admitted into the hrst group because
they fall α little short can expect α strong reaction from their financial markets. Hence,
they could plead that α vital notional interest is at stoke for them ond that EMU should
therefore be postponed until they con also join. This would require, however, α Treaty
revision, which would put the EU in an extremely difficult position.
The main problems created by variable geometry in the monetary area thus come from
the economic and political mechanisms that are amplified by financial markets. In α nut
shell, the problem is that countries that cannot participate in stage three in 1999 might
try to delay the start of EMU if they ore close to satisfying the convergence criteria. In this
sense, it is the "near periphery" that creates more of α problem than the "for periphery",
i.e. those countries that ore clearly some way from satisfying fhe convergence criteria.
In principle, all countries signed up to the conclusions of the Madrid Council of
December 1995. But important politicians in some countries hove repeatedly stated in
public that postponement should be considered. It would, however, be dangerous to
yield to α demand for postponement because there will be differences in the time required
by the "near periphery" to fulhl the convergence criteria. Establishing the principle that
everybody should wait for the ones that ore close to catching up might set in motion α
long chain of countries that were formerly far and then come close to being able to par
ticipate OS others graduate from being close to actually fulhlling the convergence criteria.
In fhe meantime, all candidates would hove fo bear the cost of the higher risk premia
that will persist until EMU really comes into existence. Moreover, it is important to estab
lish the principle that no single country should impede the others to go ahead.
81 See Gros (1995bl tor a formal model of this idea.
38
F!3 Popes
Volumel Noi 1996
Tlie t w o k e y features
of the n e w ERM are
asymmetry a n d bi la-
feralism.
Whether or not the countries that ore excluded, despite nearly fulhlling the criteria, will
suffer 0 conhdence crisis depends mainly on two factors: hrst on their determination to
press ahead with convergence efforts so that they con join in the near future, and,
secondly, on the exchange rote regime between the common currency of the core and
the countries with α derogation.
5.2. What exchange rate mechanism for the outsiders?
One of the convergence criteria in the Treaty is "the observance of fhe normal fluctua
tion margins provided for by the exchange rate mechanism of fhe European Monetary
System, for of least two years, without devaluing against the currency of any other mem
ber state" (Article 109j (1)). The subsequent article then goes on to soy that once EMU
has started, there has to be, of least every two years, on examination of the countries with
α derogation. A country con join EMU after such on examination only if it fulfils all the
convergence criteria. Hence, the Treaty implicitly seems to ossume that the exchange rote
mechanism of the EMS will continue to exist. This does not necessarily imply, how-ever,
that the EMS has to continue in exactly its present form. Since the circumstances will change
radically once fhe third stage begins, one could even argue that it has to change.
It has already been emphasised that countries that cannot participate in fhe first wove should
make clear that they will continue and perhaps even increase their convergence efforts to
be able to join EMU at the next possible dote. This would help to reduce the potential for
hnanciol market instability, but it might not be sufficient to rule out speculative attacks.
The Dublin Council of December 1996 will have confirmed the broad agreement on ERM
II that has emerged over the lost year. The two key features of the new ERM ore: i) asym
metry and ii) bilateralism (or rather "hub and spoke"). There are solid reasons for this:
i) Asymmetry. The core (even assuming it only contains only Germany, France
and some smaller countries) will be several times os large in terms of GDP and
trade as the set of countries that are outside (excluding the UK which anyway
would not be interested) (9). The difference would be even larger in terms of
the size of hnanciol markets and reputation for sfobilify. A formally symmetric
system like the EMS becomes impossible under these circumstances. In the old
ERM, Germany played α central role, although in terms of trade and GDP it
never accounted for more than 45% of the area covered by the ERM.
ii) "Hub and spokes". The trade of the periphery with the core that will represent
fhe "hub" of the system is several times larger than the trade among fhe
"spokes" (i.e. the likely outsiders). Hence the new system will not be multilateral.
91 The combined GDP of F, G, NL, A U l SF, plus IRL, LUX is about 3,000 bn ECU, compared to 800 fan ECU
for P, IT, ESP, GR, B, SW.
E IB Papers
Volumel Noi 1996 39
It will effectively be bilateral in the sense that it regulates the bilateral relation
ships between the ECB and the notional central bonks of the outsiders.
Figure 1 illustrates these point with α flow chart that assumes only α small EMU and shows
the relative size and the trade links among the major post-EMU players: the EMU 'avant
garde', the 'willing but unable' and the 'unwilling'. The group of the 'willing but unable' could
of course turn out to be much smaller, but this would only strengthen the point made here.
Figure 1. "Ins and Outs": Relative size and trade links
ff" 10 candidates GDP200BEGJ
30
Trade links in ECU billion
These consideration do not imply that each outsider should hove α different agreement.
On the contrary, fhe some type of arrangement should be offered to oil outsiders.
However, even if all outsiders ore willing to sign up to the some system it will de facto
not work in the same way for each participant. Moreover it is apparent that the nature
of ERM II will be radically different depending on whether fhe initial group is large and
membership is expected to expand quickly or whether only α small EMU starts and the
willing will take α long time to catch up. At present the first scenarios seems more likely,
but circumstances con change quickly as the experience of 1995 shows. Under the hrst
scenario, there is really no need for α fully fledged exchange rote system, since exchange
rotes are likely to be stable anyway. By contrast, under the second scenario, there would
be α real need for some system to limit exchange rote fiuctuations.
Two further points ore also cleor; First, the support from the core, i.e. the ECB, should be
linked to the convergence effort of the outsiders. The closer α country gets to satisfying
40
F;B Paoois
Volumel Noi 1996
the criteria for full EMU membership, the more support it should receive from the ECB in
defending its exchange rote if it comes under α speculative attack.
Second, there must nevertheless be α limit to the obligations of the core. The ECB could
not underwrite to α system that forces it to intervene with unlimited amounts of some pre-
specihed exchange rote because that would put its primary responsibility, namely to
keep prices stable for its members, in jeopardy. Hence, on essentially unilateral peg is
the only possible solution.
Tfie c o u n t r i e s t h a t a r e
e x c l u d e d f r o m t h e
s t a r t o f E M U , n e e d
s o m e r e a s s u r a n c e t h a t
t h e c o n s t i t u t i o n o f t h e
c o r e g r o u p is o n l y
t e m p o r a r y a n d t h a t
t h e y w i l l b e i n v i t e d t o
l o i n o n c e t h e y h a v e
c o n v e r g e d .
It is often argued that countries that remain outside will be tempted to resort to competitive
devaluations. But this fear, which is based on recent experience, seems to be unfounded. First
of oil, exchange rates ore difficult to control since they evolve with market expectations
about Riture policy. Hence, it will be difficult for any government (or national central bonk) to
"engineer" α competitive depreciation without starting α cycle of inflationary expectations
and high interest rotes that is difficult to control. Moreover, as was shown in Gros (1996α),
even large exchange rote changes tend to hove only α limited impact on unemployment
and the trade balance. Finally, since the criterion on exchange rote stability will continue
to apply, any country that is tempted by this policy would know that the price for α clearly
"beggor-thy-neighbour" policy of this type would be further delay in EMU membership.
How tight the exchange rote system will be offer stage three has started, depends thus
essentially on the country concerned. The next sub-section presents α specific proposal
that is designed to address the legitimate concerns of the excluded countries without
deloying EMU.
5 . 3 . A c o n c r e t e p r o p o s a l
The countries that ore excluded from the start of EMU, despite α desire to participate,
need some reassurance from the Union that the consfitution of the core group is only α
temporary measure and that the others will be invited to join once they hove converged.
This is certainly the intention of the Treaty. But what if initial exclusion makes con
vergence much more difficult, even if there is some exchange rote mechanism for the out
siders?
One solution might be granting α form of associate status in EMU (or rather the initial
core group). The country concerned could be invited to come under the EMU umbrella
/ 0) The government would have to declare that it accepted the obligations arising from Articles 104c(9j a n d
f 1 i } (excessive deficits procedure); 105(1},(2},{3} and (5) fmonetary policy}; 105a (notes and coins}; and
108a (empowering the ESCB}. The national central bank would also accept the obligations resulting from the
ESCB statutes (Articles 3, 9, 1 2 . 1 , 14.3, 16, 18, 19, 2 0 , 2 2 , 2 3 , 30-34 a n d 52}. However, the restrictions
specified in paragraphs 3 to 6 of Article 43 ofthe ESCB statutes would apply, (n addifion, (he country concerned
would not participate in decisions under Articles 109 (exchange rate system with rest of world} a n d I09a(2}(b}
(membership of the executive b o a r d of the ESCBj.
E iB Pooc's
Volumel N o i 1 9 9 6 41
The c o u n t r y w o u l d
g i v e u p i ts n a t i o n a l
m o n e t a r y p o l i c y , t h e
e x c h a n g e r a t e v / o u l d
b e i r r e v o c a b l y f i x e d ,
t h e p a y m e n t s s y s t e m s
u n i f i e d a n d t h e c o u n t r y
w o u l d b e s u b j e c t t o
the f u l l excessive def ic i t
p r o c e d u r e .
to benefit from lower interest rotes, but it would not participate in the management of
EMU until it hod converged in fiscal terms as well.
This arrangement could be achieved technically by α unilateral declaration that the coun
try concerned occepted oil the obligations arising from membership in EMU (10), but it
would be preferable to have α formal agreement between the EU (or rather the ECB) and
the country concerned, and the political support from fhe Council because this would
moke it much more credible with fhe markets. The agreement would specify mainly that
the notional central bonk agreed to follow the monetary policy of the ECB as if it were α
full member of the EMU. At the some time, however, if would be clear that for purposes
of decision-making in the ECB (and ECOFIN), the country would continue to be treated
OS having α derogation. In essence, the country would give up its notional monetary policy
ond replace it with that of the ECB. More precisely, this means that the exchange rote
would be irrevocably fixed, the payments systems would be unihed, actions by the ECB
would have direct effect in the country concerned and the decisions of the ECB would
hove to be applied by the notional central bonk, α portion of whose foreign exchange
reserves would be pooled in the ECB like those of the full participants. Moreover, fhe
country concerned would be subject to the full excessive deficit procedure (1 1).
All this would be officially acknowledged by the Union in conjunction with α convergence
programme outlining how the country would, with the help of lower interest payments,
satisfy the fiscal criteria by α certain deadline (1 2). Acknowledgement by both the Union
and the ECB would moke this arrangement credible and would ensure that market inter
est rates in the country concerned converge quickly to the level of the core (1 3).
This sort of associate status in EMU will deliver the benefits in terms of lower interest rotes
only if it is credible. Credibility should come already from the endorsement given by the
ECB, but it would be immensely strengthened if markets see that the exchange rote could
be defended under any circumstance. This should be the cose if one views the proposed
arrangement os α "currency board". A currency board is credible if the notional central
bonk possesses adequate foreign reserves to guarantee conversion of all of their liabilities
(the monetary base) into the common currency of the core. Would this be the cose? Box
4 below shows that most of the central bonks that might be candidates for this approach
do indeed hove enough reserves to moke the currency board approach credible.
J / ] Unilateral restricted participation would in effect be a sort o f Anschluss much like the per iod when the D M
was introduced in the territory of the former GDR. This might not be a very enticing example. In the case o f t h e
EU, however, convergence wil l take place before, not after, monetary unification. Hence, the economic diffi
culties that fol lowed German unification should not occur
12) Acknowledgement by the Union should not necessarily imply that the ECB, when setting its monetary poli
cy, would take into account economic conditions in the country that participated in the same w a y as those of
"regular" participants.
13} The actual debt service burden would decline only gradually, however, until the outstanding high interest
debt is retired. Depending on the maturity structure, this might take two years.
42
ί 'Ά Paoe'S
Volumel N o t 1 9 9 6
Box 4 . Technical conditions for the viabi l i ty of α currency board
Technically α currency board Is viable if the central bank has enough reserves to exchange all its
liabilities (the monetary base) info foreign currency. This is the case if the foreign exchange
reserves are larger than the monetary base. Column (3) in the table below shows that the ratio of
foreign assets to the monetary base is above, or close to, 1 for all countries except Italy If one
takes into account that, even in α worst-case scenario, few people will exchange their holdings of
cash, it would be sufficient for α national central bonk to have enough foreign assets to cover the
remainder of the monetary base (i.e. required reserves). Column (4) shows that this is the case by
α large margin for all the countries considered below, with the exception of Italy and Finland.
Even for these two countries, however, the shortfall is rather limited since the existing foreign
exchange reserves cover already 100-1 10% of the reserves held by bonks with the central bank.
In the Italian case, this result is due to the unusually large reserve requirements imposed on banks
in Italy. The reserve ratios in Italy are at present 5 to 10 times higher than in the rest of the EU
and have to be lowered anyway if Italy wants to participate in EMU. If reserve requirements were
only halved in Italy (leaving them much above the EU average), the international reserves of the
Banca d'Italia would be much larger than the mobile part of fhe monetary base. (Some margin is
needed since part, say 20-25%, of the foreign reserves will be pooled in the ECB.) Hence, even
the Banco d'Italia could defend the exchange rate if it previously lowered required reserve ratios
towards the EU average, provided, of course, that it does not engage in any sterilisation as done
so often in the past.
Reserves a n d M o n e t a r y Base
Belgium (bill. Franc)
Finland (bill. Mark)
Greece (bill. Drachma)
Italy (bill. Lira)
Portugal (bill. Escudo)
Spain (bill. Peseta)
Sweden (bill. Kroner)
(1) Base Money
430.4
38.0
2500.1
150.0
3001.3
7.800
163.8
(2) Foreign assets
731.7
33.5
2975.5
86.6
3706.7
6.152
175.7
Ratios based on:
Monetary base
(3) = (2)/(l)
1.7
0.9
1.2
0.6
1.2
0.8
1.1
Required reserves (4) = (2)/[(l) - cash in
circulation]
36.0
1.2
4.6
1.0
1.6
4.8
1.8
Source: International Monetary Fund
r Β i'ooc's
Volumel Noi 1996 43
All the candidates for associate membership in EMU could thus be conhdent of operating
0 tight link with the core even under α worst-cose scenario. If the markets know that there
is no chance that they con force α break in the link to the common currency of the core,
they will regard it as credible. Any country that chooses this approach could increase the
credibility even further by passing α low that obliges the notional central bonk to defend
the exchange rote through unsterilised interventions. Moreover, it is likely that if there were
really α totally unjustified speculative attack, fhe ECB would help the country concerned. If the
underlying fundamentals ore sound, credibility should thus not be α problem. If the fun
damentals ore not sound, the ECB would not recommend this approach in any event, ond
no country would (or should) dare to try it ogoinst its advice.
Technical viability is of course only α necessary, and not sufficient condition for the stability
of α currency board arrangement. The reason why central bonks usually sterilise their inter
ventions is that the large increases in interest rotes that might result if they did not are not
accepted; either because of their macroeconomic consequences (UK in 1992), or because
they could endanger the stability of the banking system (Sweden also in 1992). Central
bonks ond governments will have to convince markets that they will be willing to accept inter
est rote increases if the market tests their resolve. The experience of Belgium, which faced α
test of its commitment in 1993, shows that it is possible to present this case persuasively.
The credibility of the 'associate membership' status would also be greatly enhanced if
the country concerned followed full EMU members in converting its public debt in Euro.
This would ensure that α speculative attack would not lead to on increase in interest rotes
on public debt making self-fulfilling attacks even more unlikely.
1 do not wish to suggest that associate membership in EMU will constitute α magic wand
that eliminates all problems. But it represents an option for countries that ore very close
to qualifying for full membership. Irrevocably hxing the exchange rote with the prospect
of full participation in EMU after α couple of years is fundamentally different from hxing
exchange rotes in the environment of the 1980s (with high and variable exchange rotes)
or during the early 1990s (when some currencies were clearly overvalued). The argu
ment that experience has shown that hxing the exchange rote is impossible because
hnanciol markets could attack any exchange rote should thus not be over-rated. There
will be little reason for hnancial markets to attack on exchange rote if infiofion is low,
dehcits ore close to 3% of GDP (possibly even below), debt ratios ore declining and the
external current account indicates α good competitive position.
The political viability of this ideo depends upon its presentation. If it is characterised as
0 means of circumventing the convergence criteria, which is actually not the purpose of
this proposal, the core will veto it. The scheme merely aims to help the peripheral coun-
EIB Popes
44 Volumel Noi 1996
tries bridge the gap that separates them from the core without softening in any way the
convergence criteria for full participation in EMU.
Only countries that hove done their basic homework should be encouroged to pursue this
approach. In order to qualify, the deficit when calculated of Germon interest rotes should of
α minimum fall below 3% of GDR This discipline would also ensure that the debt ratio would
be declining once the lower interest rotes took effect. It bears reiterating that the country in
question would hove α derogation i.e. it would not be present in the decision-making organs
of the ESCB (14), which would imply that the convergence criteria hod not been suspended.
O n l y c o u n t r i e s t h a t
h a v e d o n e t h e i r bas ic
h o m e w o r k s h o u l d b e
e n c o u r a g e d t o p u r s u e
th is a p p r o a c h .
In 0 sense, Belgium and Austria hove already successfully opted for this course by pegging
unilaterally to the DM. Why can't the others follow their example? The real test of this
approach will come in the cose of ο large country, e.g. Italy. Large countries hove
always experienced more problems in acquiring sfobilify through fhe exchange rote. But
in this cose, they would not attempt to use the exchange rote to force adjustment in
prices or wages. Their problem is that they ore stuck in a low credibility-high interest rate
trap out of which it is very difficult to escape without outside help as analyzed above.
The decision to participate in EMU, even if essentially on α unilateral basis, would consti
tute one large step away from this trap. Of course, this con only facilitate adjustment. It
is in no way α substitute for fhe resolute hscol action that has to be token anyway.
6 . C o n c l u d i n g r e m a r k s
The movement towards EMU now resembles α steeple chose. The hurdles hove been set
up and governments ore now entering the finish stretch in their race towards EMU. This
paper argues that one should not overlook the infiofion criterion because the Treaty does
not provide any leeway to go over the limit of 1.5% above the three best performers and
because there is very little governments con do to infiuence infiation in the short run.
The paper also argues that one should look of changes in the debt ratio when interpreting
the fiscal criteria. It provides α concrete rule to judge what reduction in the debt/GDP ratio
'constitutes' satisfactory movement towards the reference value' as required in Article 104c.
The Council discussion in 1998 about who qualifies for the start of EMU in 1999 will
of course take political considerations into account. The countries that ore close to satis
fying the criteria but do not qualify under α 'strict' interpretation will argue that they
would pay α heavy price if they were not let into EMU and will thus press strongly for
being allowed in. This pressure could be reduced if they ore offered the alternative pre
sented in this paper, namely on 'associate' status in EMU for α one or two years to give
them the fime needed to bring their fiscal accounts under control.
/ 4) Nationals of the country concerned would also not be likely to join the Executive Board of the ECB.
EIB Pc)(x:;s
Volumel N o i 1 9 9 6 45
References
Banion, King et al. (1994). "The Infiation Tax is Likely to be Inefficient of Any Level".
Kredit und Kapital, 17, pp. 30-42.
Borro, R.J. (1995). "Infiation and Economic Growth". National Bureau for Economic
Research (NBER). Working Paper 5326.
De Grauwe, Ρ (1995). "The Economics of Convergence Towards Monetary Union in
Europe". Discussion Poper 1 17. Catholic University of Leuven.
Eichengreen, B. and von Hogen, J. (1995a). "Fiscal Policy and Monetary Union:
Federalism, Fiscal Restricfions and the No-Boilout Rule". Centre for Economic
Policy Research (CEPR). Discussion Paper 1247.
Giavazzi, E and Pagano, M. (1995). "Non-Keynesion Effects of Fiscal Policy Changes:
Internofionol Evidence and the Swedish Experience". NBER Working Paper
N o 5332.
Gros, D. (1995a), "Self-Fulfilling Public Debt Crises". Unpublished manuscript.
Centre for European Policy Studies (CEPS).
Gros, D. (1995b). "Excessive Deficits and Debts". Centre for European Policy Studies (CEPS).
Working Document 97.
Gros, D. (1996a). "Towards Economic ond Monetary Union: Problems and Prospects".
Centre for European Policy Studies. CEPS Paper No 65, January.
Gros, D. (1996b). "Germany's Stoke in Exchange Rote Stability". Centre for European
Policy Studies. CEPS Review, No 1, Summer 1996.
Gros, D. (1996c). "A Stohastic Model of SelhFulfilling Crises in Fixed Exchange Rote
Systems". Centre for European Policy Studies (CEPS). Working Document 105.
Hughes Hallett, A. and McAdam Ρ (1995). "Fiscal Deficit Reductions in Line with the
Maastricht Criteria for Monetary Union: An Empirical Analysis". Unpublished
manuscript. University of Strothclyde.
Kenen, PB. (1995). "Economic and Monetary Union in Europe. Moving Beyond
Maastricht". Cambridge University Press.
Lucas, R. (1976). "Econometric Policy Evaluations: A Critique", in K. Brunner and
A. Meltzer, eds. The Phillips Curve and Labor Markets. Amsterdam: North Holland.
Obstfeld, M. (1994). "The Logic of Currency Crises". Notional Bureau for Economic
Research (NBER). Working Paper 4 6 4 0 , February.
E. Β P:li;e::s
46 Volumel N o i 1996
Pisoni-Ferry, J. (1996). "Variable Geometry in Europe: An Economic Analysis",
in J. Frieden, D. Gros and E. Jones, eds. Towards Economic and Monetary Union:
Problems and Prospects, Oxford: Oxford University Press, forthcoming.
Pisoni-Ferry, J. and Cour P. (1995). "The Costs of Fiscal Retrenchment Revisited".
Problèmes Economiques 2448, November.
Sutherland, A. (1995). "Fiscal Crises and Aggregate Demand: Can High Public Debt
Reverse the Effects of Fiscal Policy?". Centre for Economic Policy Research (CEPR).
Discussion Paper 1246.
' • B P:;r;, .
Volumel Noi 1996 47
Appendix The accuracy of inf lat ion a n d
deficit forecasts
Inflation
A literal reading of the first article of the protocol on the inflation criterion suggests that
the examination could use the doto for the 12 months preceding the exominotion (e.g.
April 1 9 9 7 - March 1 998) instead of the average for 1997. This might be important for
the countries where inflation is still somewhat above the ceiling. But I will continue to
assume that the overage for 1997 will be used. The latest forecasts for 1997 show that
the so-called core countries (D, F, BENELUX, A) should not hove any problem in meeting
the infiation criterion. But it is apparent that oil of the four southern European countries
will have difficulties.
During 1997 α lot of attention will thus focus on price developments os well. It is there
fore useful to see to what extent forecasts for inflation hove, in the recent post, been
accurate. Table A. 1 shows past infiation forecasts for the two key core countries, France
and Germany, plus the three countries that ore likely to be 'borderline' candidates for
EMU in 1997, namely Spain, Italy and Portugal.
Table A . 1 . Inflation forecasts versus outcomes
Inflation (Price deflator private
Forecast for 1995 mode
Forecast end 94
Update mid 95
Provisional end 95
Final 96
Forecast for 1996 made
Forecast end 95
Update mid 96
Provisional end 96
Forecast for 1997
(as of end 1996)
in:
in:
consumption
E
4,5
4,9
4,9
4,6
3,9
3,6
3.6
2.9
']
IT
3,5
5,2
5,6
5,8
4,3
4,1
3.9
2.9
Ρ
4,6
4,5
4,2
4,2
3,6
3,1
3.3
3.0
Top3
2,1
1,9
1,4
1.2
2.0
1,5
1.4
1.6
Best performers
DK,D,F
B,F,NL
B,NL,SF
B,NL,SF
SF,NL,D/F
SF,D,L/S
SF,L,S
SF,F,D
Source: European Commission See table A.2. for detailed references
The forecast for infiation in 1995 that was available at the end of 1 994 was rather accu
rate for Spain and Portugal whereas Italy overshot it by more than 2 full percentage
points. There were special factors in operation in Italy in 1995, including an increase
in value added tax, which ore unlikely to repeat themselves in 1997, but this episode
4 8
E13 Papers
Volumel N o i 1 9 9 6
suggests that α substantial overshooting of the infiofion target is possible. The forecasts
for 1996 inflation for these three Mediterranean countries made at the end of 1995
hove so far proven rather accurate. But the composition of the three best performers was
different and their overage was better than expected. If there is ο similar discrepancy in
1997 between the actual performance and the forecast for the overage of the three best
performers the three countries considered here could be in serious trouble.
Fiscal deficits
It will be useful to consider briefly the accuracy of post forecasts of hscal dehcits (under
the Maastricht dehnition, i.e. for general government). Table A.2 below presents the
doto for the year 1995 for the five most important EMU candidates:
Table A.2
Forecasts for
1995 mode in:
Forecast end 94
Update mid 95
Provisional end 95
Mid 96
Forecasts for
1996 made in:
Forecast end 95
Update mid 96
Provisional end 96
Forecast for 1997
(as of end 1996)
Fiscal deficits (Maastricht defi
D E
2.4
2.1
2.9
3.5
D
2.8
3.9
4.0
D
2.9
6.0
6.0
5.9
6.2
E
4.7
4.8
4.4
E
3.0
nition)
F
4.9
4.9
5.0
5.0
F
3.9
4.2
4.0
F
3.0
IT
8.6
7.9
7.4
7.1
IT
6.0
6.3
6.6
IT
3.3
Ρ
5.8
5.6
5.4
5.4
Ρ
4.7
4.4
4.0
Ρ
2.9
Source: European Commission
Forecast end -94 from European Economy no 59
Update mid -95 from 'Statistical Annex of European Economy June 1995
Update end -95 from 'Statistical Annex of European Economy November 1995
Update mid -96 from 'Statistical Annex of European Economy June 1996
Update end -96 from 'Statistical Annex of European Economy November 1996
Volumel Noi 1996 49
Table A.2 suggests that in the post the forecasts hove been rather accurate, with two
exceptions for 1995. Germany performed much worse than expected (also in 1996)
whereas Italy performed much better. Given that the Federal government controls only
about one half of all general government spending in Germany this is easily explained.
But this implies that it will be difhcult for the German Federal government to achieve the
3% for general government that constitutes the Maastricht criterion.
All the forecasts ore always subject to the underlying assumption concerning economic
growth and ore based on unchanged policies. This basis was apparently useful in the
post. It remains to be seen whether the various occounfing movements proposed by the
French and Italian governments will actually be sufficient to achieve the 3% dehcit target
OS planned.
i s Pai>e's
50 Volumel Noi 1996
On the feosibility of EMU vsfith diverse European economic
transmission mechanisms
ole Rummel
Economist, Chief Economist's Department
The question arises
infhether changes in
Interest rates w i l l have
similar effects on the
financial systems of
each country in the
union? This problem is
generally val id for a l l
kinds of macroecon
omie variables.
1. Overv iew
It is by now widely expected that not oil European Union jEU) member countries will join
the third and hnol stage of European Monetary Unification (EMU) on 1 January 1999,
either because countries ore unwilling to replace their own national currencies with the
Euro, or because they are unable to meet the entrance requirements as laid down in the
Maastricht Treaty.
The formation of EMU raises several formidable questions. Of interest is thus not only the
question of who is going to join EMU, but also what is going to happen to those coun
tries who are either unwilling or unable to join. In many respects, the latter may be regar
ded as α more challenging and far-reaching question because of its strategic implications.
In the words of Thygesen (1995, p. 7), "there is clearly α strategic interdependence of
the decision to join and the nature of the arrangements with the rest". While membership
in EMU from the start presupposes convergence of economic performance and structures
among members, the some cannot be sold about those countries left outside the union.
As we will see below, the "outs" will form α heterogeneous group largely by default,
where some countries will choose to be "out" because of their political opposition to the
loss of sovereignty associated with EMU, while others will be "out" because of important
differences in their respective convergence efforts. By their exclusion, the "outs" should
not be in α position to - intentionally or unintentionally - jeopardise the achievements of
the Single Market through exchange-rate instability or otherwise, and the relationship be
tween the "ins" and the "outs" thus becomes of vital importance. (1)
The future monetary policy under EMU will entail the use of one interest-rate policy for
the whole of the monetary union. In particular, the question now arises whether changes
in interest rotes controlled by the European Central Bonk (ECB) will hove similar effects
on the hnanciol systems of each country in the union? Much more than that, though, this
problem is generally valid for all kinds of macroeconomic variables. The analysis of the
symmetry and correlation of macroeconomic shocks hitting European economies, which
has hitherto been used to differentiate European economies into "ins" and "outs," is
sadly lacking in this aspect. Even if shocks hitting European economies were symmetric,
in the sense that they affect each country in the some way, there is still no guarantee that
the responses to these shocks ore symmetric as well. The upshot of this analysis, name
ly on invesfigotion of the transmission mechanisms of European economies, has re
ceived little attention in the literature. (2) This is unfortunate, as the transmission mech
anism is not only another instrument to potentially differentiate "ins" and "outs," but also
0 possible stumbling block in the transition to, and even future functioning of, EMU.
1) The question of access of non-EMU countries to the future European payments system, TARGET, may be a
case in point.
2) Notable exceptions are the two studies by Ballabriga et al. (1993, 1995).
E 'Β PoDC'b
Volumel Noi 1996 51
The remainder of the paper is set out as follows. The next section outlines the founda
tions for the creation of α monetary union, which is commonly associated with the idea
of an optimal currency area (OCA), i.e. α union of countries in which the circulation and
use of α single currency would be optimal. Two criteria for α successful OCA ore brief
ly investigated empirically in Section Three, while the subsequent Section focuses on the
main aspect which may complicate the analysis of OCA's, the transmission mechanism
of European economies. Three previous studies of the transmission mechanism ore also
briefiy reviewed, before the hnal Sections present the empirical results for the degree of
asymmetry of transmission mechanisms in terms of partial correlofion coefficients, fore
cast error variance decompositions and impulse response functions. As the OCA argu
ment relies on the incidence of highly posifively correlated shocks of similar size, the
asymmetric transmission mechanism of symmetric shocks within European economies, if
it exists, may weaken the posifive evidence for the future feasibility of EMU.
2 . The F o u n d a t i o n s f o r t h e t h e o r y o f O p t i m a l C u r r e n c y A r e a s
The biggest change occurring under α currency union such os EMU will be the "loss of
the exchange rote as on instrument of macroeconomic adjustment (von Hogen, 1993,
p. 264)". The literature tells us that α currency union will be desiroble if the inability to
change the value of the internal exchange rate does not moke the elimination of dise-
quilibrio more difficult or more costly. Mundell (1961) showed that α single currency
would reduce tronsocfions and information costs and the degree of imperfect competi
tion; but that having hxed exchange rates would prove costly in the face of asymmetric
shocks or price rigidities. However, those costs would be smaller if labour and capital
were sufficiently mobile. McKinnon (1963) argued that openness to trade was on import
ant criterion because changes in exchange rotes would, in that cose, hove few beneh
ciol effects in terms of real wages or terms of trode. Finally, Kenen (1969) has argued
that countries with α high product diversificafion would find it easier to maintoin α cur
rency union since they would be less vulnerable to, and hence better able to absorb,
shocks to particular industries or sectors. Von Hagen (1993, p. 264) concluded that the
costs of monetary union "will be small, if the size of transitory, country-specihc shocks
within the union is small compared to common shocks and permanent shocks, if capital
and lobor mobility ore high within the union, ond if the majority of the members' trade
occurs with other members of the monetary union".
In short, α currency union may be desirable if (o) the countries ore mutuolly open to
trade; (b) producfion factors ore mobile; (c) wages and prices are flexible in both direc
tions; (d) there is α high degree of product diversificafion; and, (e) the countries con
cerned do not display signihcont asymmetries in shocks and structures.
E IR Poiîo's
52 Volumel Noi 1996
A n i m p o r t a n t w a y o f
assessing the f e a s i b i l i t y
o f m o n e t a r y u n i o n is
f o t r y a n d g a u g e t h e
d e g r e e t o vvhich shocks
h i t t i n g E u r o p e a n e c o n
o m i e s a r e a s y m m e t r i c
i n n a t u r e .
A number of recent papers hove reviewed these conditions and the extent to which they
ore sofisfied in Europe. (3) The conclusion that the EU economies ore sufficiently open
to trade among themselves and that capital is highly mobile appears uncontroversiol.
Likewise, the conclusion that labour is largely immobile for linguistic and cultural rea
sons, OS well OS the personal and social costs of migration, is accepted by oil.
Provided that labour is immobile and that wage and non-wage labour remain infiexible
in Europe, then asymmetric shocks will lead to signihcont and persistent disequilibrio
under α common currency. The conclusion of whether Europe qualihes os an optimal cur
rency area or not thus depends on the balance of benehts against costs, where the lat
ter ore related to the asymmetry between national shocks or transmission mechanisms.
Hence, to test whether Europe - or α subset thereof - could function well enough os α
single currency area requires the exominofion of the symmetry of shocks ond their trans
mission.
3 . A n e m p i r i c a l a s s e s s m e n t o f c u r r e n c y u n i o n m e m b e r s h i p c r i t e r i a
There is no single criterion that con be used to assess the viability of ο currency union
between EU member states. For that reason, increosing attention has turned to the ana
lysis of shocks hitting European economies since shock-obsorption combines the net
infiuence of several of the traditional criteria. An important way of assessing the feasi
bility of monetary union is then to try ond gauge the degree to which shocks hitting
European economies ore asymmetric in nature. The more integrated economies ore, the
smaller the likelihood of asymmetric shocks, i.e. shocks hitting one country or region
worse than others.
This line of investigofion is most closely associated in the empirical literature with the
seminal studies by Boyoumi and Eichengreen (1992a,b), who calculated the sizes and
correlaHons of demand and supply shocks hitting European economies. The methodol
ogy distinguished between demand shocks, which were assumed to be transitory in
nature and refiected monetary and/or fiscal policy, and supply shocks, which had per
manent effects and were independent of policy. The main hnding of the authors was that
Germany and its immediate neighbours formed α distinct subgroup within EU econ
omies. (4) In other words, ο Core EMU was deemed economically feasible, whereas α
wider EMU of oil EU countries would not be optimal.
The analysis proceeded in two steps, one looking at the sizes of the demand and supply
shocks and the other at the degree of correlation between them. Table 1 shows the esti
mated sizes of the shocks in order to evaluate the first condition for their symmetry.
3) Masson and Taylor (1993) o n d Tavlas (1993) provide general surveys.
4} It has to be said that the work of Boyoumi and Eichengreen is not without its critics. See, for example,
Bofinger (1994) a n d Demertzis, Hughes Hallett and Rummel (1996).
;:B P;ir::-s
Volumel N o i 1 9 9 6 5 3
Table 1. Sizes of Supply and demand shocks in percent, 1961 -1988
Supply Shock
Austria
France
Belgium
Sweden
Denmark
Germany
Netherlands
Finland
Ireland
Italy
Spain
UK
Portugal
Greece
1.2
1.2
1.5
1.6
1.7
1.7
1.7
2.1
2.1
2.2
2.2
2.6
2.9
3.0
Demand Shock
Sweden
France
Germany
Netherlands
Spain
Belgium
Greece
Austria
UK
Italy
Denmark
Finland
Portugal
Ireland
1.0
1.2
1.4
1.5
1.5
1.6
1.6
1.7
1.7
2.0
2.1
2.4
2.8
3.4
Countries are arranged in ascending order ofthe standard deviation ofthe respective shock.
Source: Boyoumi and Eichengreen (1992b], Table 3.
We con see that the countries with the six smallest shocks (Austria, France, Belgium,
Sweden, the Netherlands, Germany and Denmark for supply shocks; Sweden, France,
Germany, Spain, the Netherionds, Belgium, Austria and the UK for demand shocks)
show α large degree of overlap, from which the authors concluded that several coun
tries on both lists may be candidates for α monetary union. The evidence presented by
Bayoumi and Eichengreen becomes even clearer when the correlations of the shocks ore
introduced, which ore displayed in Figure 1. The theory prescribes that countries form
ing α monetary union should hove highly (posifively) correlated shocks of similar magni
tude. In other words, countries would be in the top right-hand corner of the diogrom with
both high demand and supply correlations. Germany os the biggest European economy
was chosen os the benchmark for the correlofion coefficients.
E 'Β Pacers
54 Volumel N o i 1996
Figure 1. Correlatìons of demand and supply shocks with those of Germany, 1961 -1 988
Q
-a : . ο
•0.1 Ireland •
Greece
• UK
0 0.1 Supply Shocks
•
Portugal
Italy • «Sweden
Finland
0.2 0.3
• Spain
• Austria
0.4
France»
0.5
Denmark • Netherlands
• Belgium
0.6 O.i
Source: Bayoumi and Eichengreen (1992b), Chart 5
Figure 1 shows that there exists α distinct grouping of Austrio, Denmark, France, the
Netherlands ond - possibly - Belgium, from which the authors concluded that Germany
and some of its immediate neighbours have the makings of α natural OCA.
The analysis con be extended by looking at the correlafions and sizes of each shock
individually. Here, the aim is to hove small, but highly correlated shocks. For supply
shocks, the some five countries fall into the area previously identihed os being optimal
for monetary union, i.e. Austria, Belgium, Denmark, France and the Netherlands. This
result is borne out by the analysis on the demand side, for which Austria, Denmark,
France and the Netherlands emerge as prime candidates for α monetary union with
Germany.
Altogether, Bayoumi and Eichengreen's conclusion about "ins" and "outs" has identified
Germany and some of its immediate neighbours os forming the nucleus of the "ins".
Naturally, highly correlated shocks of similar size in the post ore no guarantee of
unchanged shocks in the future. Bofinger (1994), for example, has raised the fear that
EMU will entail ο regime break, in which case the predictive power of observed shocks
before the formation of EMU may not necessarily carry over into the EMU-regime.
4 . The t r a n s m i s s i o n m e c h a n i s m o f e u r o p e a n e c o n o m i e s
The last section has left us with some support for the hypothesis that α subset of European
countries, mainly composed of Germany and her immediate neighbours, fulfils the re
quirements for on optimal currency area. Even though the shocks are not perfectly cor
related, in the sense that none of the correlofion coefficients between the shocks hitting
the European economies ore equal to one, there is little to suggest that the correlations
ore in fact too low not to warrant the existence and functioning of on OCA.
; 'B Pooi'-s
Volumel Noi 1996 55
The trodifionol argument in favour of OCA's involved the analysis of asymmetric (real
demand) shocks. But EU member states ore marked by a high degree of producfion diver
sification, which may well make these countries immune to major asymmetric shocks.
Bofinger (1994, p. 29) thus concludes that "it is fundamentally difficult to imagine α major
asymmetric real demand shock at oil. In fact, in the many publications of the EMU critics
not α single concrete example con be found". (5) Similarly, the Commission of the
European Communifies (1990) predicts that "Economic integrofion will moke the occur
rence of country-specific shocks less likely ... (p. 136)" and that "asymmetric shocks in
the Community, even though they exist, are likely to diminish ... (p. 147)".
The next secfion will turn to the investigation of the occurrence and propagation of the
largely symmetric shocks to Core-countries identified by Bayoumi and Eichengreen
(1 992a,b) (6), and take the analysis one step further. Common shocks, be they external
or sector-specihc, con hove asymmetric consequences depending on the differences
among countries in initial situations, economic structures and behaviour or preferences.
The focus thus turns to the investigation of the transmission mechanisms of notional
European economies, for which the rationale is given by Bayoumi ond Eichengreen
(1992, p. 2):
The costs [of monetary union] are ossocioted with the need for the members of
0 monetary union to run identical monetory and similar hscal policies. The
weight that should be attached to this imperative depends on, among other
things, the incidence of shocks. If disturbances are distributed symmetrically
across countries, symmetrical policy responses will sufhce.
The aim of the invesfigotion is thus not α detailed modelling of European economies, but
α representofion of the transmission of α common external shock in order to assess its
degree of symmetry.
In general, the approach token in this study is reminiscent in style to Lostropes and Koroy
(1990), who investigated the transmission of aggregate shocks between the US and
three major European countries, namely France, Germany and the UK. The analysis
extended over both the fixed and the flexible exchange-rate period, and concluded that
fiexible exchonge rotes do not, in general, provide α complete insulofion from foreign
shocks. Furthermore, the degree of insulation and interdependence differed substan
tially across European countries.
5) A judgement o f the viability of an O C A is mainly based on the future occurrence of asymmetric shocks.
While it is not possible to claim that asymmetric shocks no longer exist, it is difficult to foresee future idiosyn
cratic shocks such os German unification.
6} Members of the Core are Belgium, Denmark, France, Germany a n d the Netherlands, while the Periphery
encompasses Greece, Ireland, Italy, Portugal, Spain a n d the UK.
EIB PODCIS
5 6 Voluirel N o i 1 9 9 6
A r e E u r o p e a n e c o n
o m i e s p r o n e t o d i f f e r
e n t t r a n s m i s s i o n m e c h
a n i s m s o f s y m m e t r i c
s h o c k s , t h u s l e a d i n g
t o a s y m m e t r i c e f f e c t s ?
Of even greater interest for Europe ore the studies by Bollobrigo et al. (1993, 1995),
who looked at the propagation of real and nominal shocks among four European eco
nomies. The four countries investigated, namely France, Germany, Spain ond the UK,
encompass not only 70% of the EU's total GDP, but also form α rather heterogeneous
group. As such, the group includes two Core and two Periphery countries. Each model
contains three variables (the rote of change of the oil price, US output growth and the
US nominal interest rote) emonofing from the "rest of the world" in addifion to three
variables each from the four European countries (real output growth, infiofion and the
nominal interest rote). The authors were thus able to additionally investigate European
interdependencies. Employing α Boyesion VAR (7) analysis with quarterly doto from
1970 to 1993 to characterise the responses to common and specihc, nominal and real,
shocks in the four European economies, Bollobrigo et al. found that asymmetric shocks
hove dominated in the short-run, which runs counter to the Bayoumi and Eichengreen
(1992a,b) result of α symmetric nature of shocks. Importont asymmetries existed on the
real side over the whole sample period. Most notably, the weight of external shocks was
different across countries. Some aspects of asymmetry were even higher with respect to
infiofion, while α high degree of symmetry existed on the financial side, refiecting no
doubt the process of European financial integration.
5 . E m p i r i c a l e s t i m a t i o n r e s u l t s : P a r t i a l c o r r e l a t i o n coef f ic ients
By looking at the impact of two real and two nominal shocks at the European level on
twelve EU member countries, this study will try to complement the earlier invesfigofions
by Ballobrigo et al. (1993, 1995). The vector outoregression (VAR) methodology pion
eered by Sims (1980) con help us to address the questions idenfified above, namely ore
European economies prone to different transmission mechanisms of symmetric shocks,
thus leading to asymmetric effects. (8) In total, the VAR system for each country includes
eight variables in levels, more specihcolly the narrow money supply (M), the long-term
government bond yield (LR), the consumer price index (CPI), the industrial producfion
index (IP) and the equivalent European aggregotes of the foregoing. (9) The four
variables will be used to ossess monetary, hnanciol, nominal and real integrofion of EU
economies respectively. (10) Overall, the model will invesfigate the relative importance
of external (European) and internal (domesfic), as well os real (industrial producfion) and
nominal (CPI), shocks. The European money supply and consumer price index data were
7) Appendix Β contains a more detailed description of Bayesian VAR modelling.
8} The VAR methodology lends itself to the purpose of estimating dynamic systems.
9} Ireland h a d to be excluded from the investigation as it does not produce a monthly CPI series. The danger
of a break in the German money supply series due to unihcation was circumvented by scaling al l monthly
hgures prior to fhe date upwards by 12%, which was roughly the increase in the money supply M 1 upon uni
hcation.
10} The analysis by Lostropes a n d Korey ( 1990) included bilateral exchange rates against the US Dollar in the
VAR's. For the purposes of this study, however, it was decided to follow the example of Ballabriga et al. (1993,
1995} and not include exchange rates in the system, which may perhaps give rise to an omitted variable problem.
Volumel N o i 1 9 9 6 5 7
token from the Main Economic Indicators of the OECD, while α European industrial pro
duction index and the long-term interest rate were constructed by the author. (11) Each
variable in the VAR system is treated as on endogenous variable and is modelled os α
linear function of the post realisations of all the variables and an unpredictable error
term. A vector of variables, x , , con thus be written as:
[1] X, = ^ A,x,_,-He,
/=i
where, if x , is α vector of η variables. A, will be on (w χ η) matrix of coefficients, lì is
the log-length of the VAR and e, is α vector of normally distributed errors. Each variable
in X, is thus regressed on lagged values of both itself and oil the other variables in the
system. With α log length of ^ and η variables, the number of coefficients to be esfi-
moted in each equation is equal to n l i . For the whole system with η equations, this num
ber increases to w χ [nli]. However, the number of observations typically available is
inadequate for estimating the coefficients of the VAR with precision. The likely over-poro-
meterisofion of the VAR therefore led us to take recourse to Boyesion VAR modelling,
where prior beliefs about the distribution of the coefficients ore explicitly included in the
estimofion process. Bayesian modelling allows us to circumvent the problem of the over-
porometerisotion of the VAR and is explained in more detail in Appendix B.
Partial Correlation Coefficients of fhe Variables in the System
It is instructive for the analysis which follows to first look at the porfial correlation coef-
hcients between the residuals in the different systems. Their importance lies in the fact
that they allow us to gouge some inifiol economic interactions between the domesfic and
the European voriobles. The partial correlotion coefficients for the different VAR systems
are given in Table 2.
/ /] The EU industrial production index is a weighted average of the national production indices, with weights
equal to the share of the respective national GDP relative to total EU GDP. Following convention, the long-term
interest rate for the EU is a weighted average of the national long-term interest rates, with weights equal to the
share of the constituent currencies in the dehnition of the ECU basket.
i 'B •'apc's
58 Volumel Noi 1996
Table 2. Significant porfial correlofion coefficients of esfimated residuals
Danish System Belgian System
EULR EUCPI EUlP DKM EULR EUCPI EUlP
EUCPI * BGLR
DKM * BGCPI
DKLR * * BGIP
DKCPI
DKIP
French System Dutch System
EUM EULR EUCPI EUlP EULR EUCPI EUlP
EUCPI * * NLLR
EUlP NLCPI
FRM * NLIP
FRLR
FRCPI
FRIP
German System Italian System
EUM EULR EUCPI EUlP EUM EULR EUCPI EUlP ITCPI
EULR * * EUCPI
BOM * * ITLR
BDLR * * * * ITCPI
BDCPI * * ITIP * • . *
BDIP
Greek System Spanish System
EUM EULR EUCPI EUlP EULR EUCPI EUlP ESM ESLR
EUCPI * EUCPI *
GRIP * * * ESLR
ESCPI
ESIP
Portuguese System UK System
EULR EUCPI EUlP PTM EUM EULR EUCPI EUlP UKLR
EUCPI * EUCPI
PTM * UKLR
PTCPI * * * UKCPI
PTIP * * UKIP
A * (**) indicates significance at the 10% (5%) level. The first two letters denote the country
(BG = Belgium, DK = Denmark, FR = France, BD = Germany, GR = Greece, IT = Italy,
NL = Netherlands, PT = Portugal, ES = Spain, UK = United Kingdom, EU = European Union);
while the remaining letters denote the vorioble (M = money supply, IP = industrial production,
LR = long-term government bond yield, CPI = consumer price index).
[ • : i • ' a i y - s
Volumel Noi 1996 59
It is perhaps not surprising that the European money supply measure is significantly corre
lated with the long-term government bond yield in three of the four largest members of the
EU. Between them, France, Germany, Italy and the UK moke up almost 80% of EU total
GNP on the basis of current ECU GNP weights. For France, it is not the long-term interest
rote which is (negofively) correlated with European money, but the French money supply.
The fact that the European long-term interest rote is only significantly (positively) correlated
with the European money supply variable in the Germon system may lend some credence
to the hypothesis that Germany effectively sets long-term interest rates for the rest of Europe.
The r e s t o f t h e c o r r e l a
t i o n s shovi/ t h a t n o n e
o f t h e E u r o p e a n c o u n
tr ies a r e p a r t i c u l a r l y
w e l l i n s u l a t e d a g a i n s t
c o n t e m p o r a n e o u s
s h o c k s .
The rest of the correlations show that none of the European countries are particularly well
insulated against contemporaneous shocks. As such, European interest rates ore - per
haps surprisingly - significantly negofively correlated with oil domesfic long-term interest
rates except for Greece and Portugal. The existence of signihcont correlotions demon
strates capital links between the European countries. Their absence for Greece and
Portugal may be explained either by capital controls, or interest rote torgefing by the re
spective central bonks. In fact, Poriugal only lifted capital controls in 1990, while
Greece still maintains low-level ones. This may give these countries some degree of
contemporaneous policy independence. But the European interest rote not only offects
the domestic interest rotes, it is also - with the exception of Belgium, Germony and the
Netherlands - significantly negatively correlated with the European CPI in the remaining
notional systems, indicofing signihcont changes in real long-term interest rates.
Furthermore, there is evidence of correlofion between the European long-term interest
rate and the notional money supplies of Denmark and Portugal.
Altogether, this con be token os comforting evidence to suggest that α European Union-
wide money demand function con be used for α pon-Europeon monetary policy. At the
some time, however, the evidence presented by the parfiol correlofion coefficients tells
us little about the symmetry of the transmission mechanisms as such. The partial corre
lofion coefficient between European and notional interest rotes range from on insignifi
cant value for Portugal to α highly significant-0.83 for Germany, and α grouping of the
six countries with the highest portiol correlation coefhcients, namely Germany, France,
the Netherlands, Belgium, the UK and Spain, only contains four of what Bayoumi and
Eichengreen (1992a,b) classify as Core countries.
European industrial producfion, finally, has α pervasive infiuence among oil European
countries, even though it is - again surprisingly - signihcontly (negofively) correlated
with oil eleven nafionol output levels. There is thus α dehnite interrelofionship on the real
side. In Belgium, European industrial production is also correlated with the long-term
interest rote, while in Germony α statistical relationship exists with the money supply. In
Italy, this relofionship also extends to the price level.
60 E13 Pof.t-s Volumel Noi 1996
All in oll, links between countries con be found on the output side, which lends some
suppori to the fact that the strength of the links may not only lie on the nominal, but on
the real side os well.
6 . E m p i r i c a l e s t i m a t i o n r e s u l t s : V a r i a n c e d e c o m p o s i t i o n s a n d i m p u l s e
r e s p o n s e f u n c t i o n s
The analysis of EMU pre-supposes convergence in infiation, interest rotes, public debt
and dehcits os EMU starts, but little attention is being paid to α congruence in monet
ary behaviour (among others) os EMU proceeds. The potentially different economic
structures of EU economies will be the focus of the following analysis, where the VAR
methodology is employed to study the dynamics of the system. The plethora of derived
empirical information con be conveniently summarised with the help of forecast error
variance decomposifions and impulse response functions (IRF's).
These meosures will help us to evaluate the two conditions for "symmetric shocks," name
ly 0 common origin and the symmetric transmission pattern both in terms of size and per
sistence. In response to α common shock to α European variable, the variance decom
position will quanfify the relative importance of each external shock, while the impulse
response function explains its transmission mechanism. The former thus gives us on indi
cation of the degree to which national economies ore "open" to shocks at the European
oggregote level. An asymmetric shock in the variance decomposifion is thus one which
emerges as α relevant source of variability in the domesfic variables for a subset of coun
tries only. The IRF, on the other hand, allows us to characterise symmetries in the trans
mission of these shocks in terms of the sign of the disturbance, its magnitude and its per
sistence. An idiosyncrofic shock from the variance decomposition must thus be borne out
by the evidence in the IRF's os well.
Forecast error variance decomposifions break down the variance of the forecast error
for each variable into components that con be attributed to each of the endogenous
variables and report the proportion of the movement of α variable due to its "own" shock
versus shocks to the other variables. Table A in Appendix A shows the variance decom
posifions for the responses of the four dependent notional variables.
The porficulor ordering of the variables in Table A, i.e. money supply (M) >-long-term
government bond yield (LR) >-consumer price index (CPI) rindustriol production (IP), was
derived by imposing α "semi-structural" interpretation on the model. This causal - and
temporal - ordering means that movements in money precede movements in the other
three variables, i.e. interest rotes, the price level and output. In addition, aggregate
European variables appear before national European variables. This restricfion results
Volumel Noi 1996 61
from the underlying assumpfion that Europe os α whole is α much larger economy than
any of the consfituent ports, from which it follows that the European oggregote is less
likely to respond contemporaneously to shocks to nafionol European economies. On the
other hand, the structure we impose on the system does not rule out effects that lagged
notional variables may hove on the aggregate European ones, and allows us to inter
pret EU innovations as fundamental symmetric shocks with regard to the nafionol
European variables in the system.
The results from the variance decomposifions con be summarised os follows:
Monetary Integration
Regarding the importance of the external European monetary shock, we find thot - with
the excepfion of Italy and the UK, for which the weight of the European shock is above
20% and 7% respectively - oil countries ore relatively "closed" to monetary shocks ot
the aggregate European level. Except for Italy, the percentage of variance explained by
the own shock is above 70% for oil countries. (12) With α few excepfions, the three
remaining European variables (European long-term interest rote, European CPI ond
European industrial production) ore in general more important in explaining the fiuc
tuations in the domesfic money supply than the European money supply shock. Of the
four European variables, it is mainly the European long-term interest rote which has the
second most important infiuence on the domestic money supply.
Financial Integration
In terms of financial integration, esfimated by the importance of the European long-term
interest rote shock, we con identify α group consisfing of Germany and France with per
centages around 60%, the Netherlands with 4 2 % and another group mode up of
Belgium, Italy and the UK with percentage weights in the twenties. In general, the clos-
sificafion given above is consistent with the important degree of financial integration in
the EU, and mirrors the hnanciol linkages between France, Germany ond - to α lesser
extent - the Netherlands.
Nominal Integration
The analysis of nominal integrofion in the variance decompositions reveals α group with
α weight of the European CPI shock of over 20%, consisfing of Belgium, Spain and the
UK; four countries with weights between 10% and 20% (France, Germany, the
Netherionds and Portugal) and the remaining three countries, for which the variance
decomposifion is less than 10%. Most of the variability of the CPI is therefore explained
by the domesfic variable. In general, the interocfion on the nominal side is smaller than
that on the real side. With the exception of the Netherlands, Portugal, Spain and the UK,
ί 2} Italy's monetary openness on the basis of the variance decompositions is surprising. A further investigation
is beyond the scope of this paper, however.
•'. 3 Por^r s
6 2 Volumel N o i 1 9 9 6
the percentage of variance explained by the nominal shock exceeds that of the real
European shock examined in more detail below. Other European variables enter with per
centages equal to or exceeding that of the European nominal shock. Examples for the lat
ter ore the infiuence of the Europeon long-term interest rote on the variability of the French,
German, Italian and Dutch CPI. The transmission of α European real shock on the domes
tic CPI level is generally small, in the sense of being less than 5%, even though it seems
to be quite important for Spain, for which the percentage lies above 16%. (13)
Real Integration
Finally, we turn to the discussion of real integrofion via the European industrial produc
fion variable. Here, α clear distinction emerges between α group of countries with per
centages above 20%, composed of Belgium, France, Germany, Italy ond Spain, and the
rest of the EU countries, which have percentages less than 20%. The latter thus reveal α
small degree of "real openness," as indicated by the large percentage explained by the
own industrial production shock, which is obove 70% in every cose. As the weight of
external real shocks is different across countries, the evidence points to the fact that
asymmetric industrial producfion shocks ore important. However, the asymmetric real
shocks identìhed ore not limited to one porticulor type of country. In terms of their explana
tory power, no other European aggregate variable is important in explaining domesfic
industrial producfion variability. The one exception is the infiuence of the Europeon long-
term interest rate on German industrial production with α figure of 1 1%.
Furthermore, there does seem to be α real versus nominal dichotomy, in the sense that
neither domestic nor European nominal shocks, i.e. to the CPI, ore important in explain
ing real variables, i.e. industrial production.
It now remains to be seen whether the results on the origin of the shocks is borne out by
their propagation and transmission, which is analysed with the help of impulse response
functions. The latter ore useful in tracing out the response over time of on endogenous
variable, i.e. one that is determined within the VAR system, to α one-period shock -
hence the "impulse" - in that and in every other endogenous variable. For example,
shocking one (endogenous) variable in the VAR will affect oil the other (endogenous)
variables, as the shock filters through the model. The IRF's then visually represent the fime
path of the effects of the shocks on the variables contained in the VAR.
In our cose, we ore dealing with α system of eight equafions in the four European and
four notional variables for each country. As all variables ore endogenous, i.e. determined
within the model, α shock to any one of the variables will affect oil the other variables
in the VAR. This allows us to examine the responses of the respecfive nafionol variables
to α one standard deviation shock in the European money supply, the European long-
13} For Belgium, France, Greece, Italy, Portugal and the UK, the percentage falls between 5% and 10%.
• ' 4 ^ • . • . • • •
Volumel Noi 1996 63
term interest rate, the European consumer price level and European industrial produc
fion. The approach token in this context is to look at the effects of aggregate European
shocks, as compared to the effects of notional shocks, on the countries of the EU over
the period of the EMS, in order to see how similar the responses ore. (14)
I f the asymmetric
effects of shocks are
large enough, a mon
etary union w i l l be
come extremely d i f f i
cult to operate for
exactly the same rea
son as it w o u l d have
been under asymmet
ric shocks.
Symmetric shocks con generate persistent disequilibria if the degree of labour mobility
or wage and price fiexibility varies between countries. By contrast, product diversihco-
fion would moke countries appear more symmetric since either industry or country-spe
cihc shocks would become rather small, or the time profile of the transmission mech
anisms would become more similar because shocks would be obsorbed equally rapidly
everywhere. Conversely, regional specialisation - α noturol consequence of the oppor
tunities for exploiting the scale economies and wider ranges of comparative advantages
which the single European market now offers - would imply asymmetric transmission
mechanisms and disturbances that were more persistent. If the asymmetric effects of
shocks ore large enough, α monetary union will become extremely difficult to operate
for exactly the some reason as it would have been under asymmetric shocks. (15)
The impulse response functions in the chief target variables (money supply, long-term
interest rates and the CPI level) over α period of 60 months - divided into one diagram
each for Core and Periphery countries - are given below in Figures 2 through 4. In order
to facilitate comparison, countries ore divided into Core and Periphery according to the
Bayoumi and Eichengreen (1992a,bl classification, i.e. in each Figure, the top panel
shows the IRF's for the Core countries Belgium, Denmark, France, Germany and the
Netherlands; while the bottom panel displays the IRF's for Greece, Ireland, Italy,
Poriugal, Spain ond the UK. In the diagrams, the horizontal axis measures the length,
and the verficol axis the size, of the response. The similarity of responses will be assessed
in comparison to Germany, which, being the largest economy in Europe, makes it the
obvious standard for comparison.
14) The extensive results for the domestic transmission effects of national shocks are available from the author
upon request.
15} A prime example of f iow European economies can differ institutionally is given by the UK. It is often alle
g e d that, because of Ihe prevalence of short-term personal borrowing a n d floating-rate mortgages, the UK dif
fers from its continental partners in its greater sensitivity to short-term interest rates.
6 4
Î . B P . H ; , . : .
Volumel No i 1996
Figure 2. Impulse response functions of money supplies to α one standard deviation shock in
the European money supply over α period of 60 months
0.003
-0.003 __.-
-0.004
0.003
0.002
sGRE
•0.001
•0.002
-0.003
There is some similarity between Germony and the Benelux countries, as oil effects peak
within one year. Denmark and France, on the other hand, either hove α much larger re
sponse (Denmark) or α much later maximum (France). The Periphery generally shows
very late minima and moximo and no similarity in the respective responses. At the some
time, Italy and the UK have IRF's which are not all that different from those of Germony
and the Benelux.
EIB Papers
Volumel N o i 1996 65
Figure 3 displays the responses to α one standard deviation shock in the European long-
term interest rate.
Figure 3. Impulse response functions of long-term interest rotes to α one standard devia
tion shock in the European long-term interest rote over α period of 60 months
The r e s p o n s e o f t h e
UK is m o r e c o m p a r a b l e
t o t h a t o f G e r m a n y ,
a n d p o i n t s t o a r a t h e r
a m b i g u o u s p o s i t i o n o f
t h e U K b e t w e e n t h e
Core a n d the P e r i p h e r y
o n t h e n o m i n a l s i d e .
0.004
0.0035
0.003
0.002
0.0015
0.001
0.0005
-f-L4f-t
-0.0005 0 6
-O.OOI
The IRF's of Germany, Belgium and the Netherlands peak within three months and hove posi
tive effects of less than three years, offer which the responses to the shock turn negative. While
France shares the maximum impact of the shock after three months, the response to the shock
lasts much longer. Denmark, on the other hand, shows α maximum after six months only.
Again, the Core emerges with respect to Germany and her smaller neighbours, and may
66
EIB Prip».· Volumel Noi 1996
not include Denmark and France in this cose. Most of the countries in the bottom pone! show
ο maximum effect of the shock long after Germany does, and their IRF's do not cross the
horizontal axis until year four. Spoin and the UK stond out os their IRF's peok immediotely
and then die down, even though Spain's initial impact is much larger. Once again, then,
the response of the UK is more comparable to that of Germany, and points to α rother om-
biguous position of the UK between the Core and the Periphery on the nominal side.
We can now look at the effects of α one standard deviation shock in the European CPI
index, for which Figure 4 presents the results.
Figure 4. Impulse response funcfions of consumer price indices to α one standard devio-
fion shock in the European consumer price index over α period of 60 months
0.0025
0.002
0.0015
0.001
0.0005
0 6 12 18 24 30 3S^~-~__42 48 " " - * • - -
•0.0005
0.0025
0.002
0.0015 —
0.001
0.0005
•0.0005
EIB Papers
Volumel Noi 1996 67
A l o n g - t e r m i n t e r e s t
s h o c k l e a d s t o a clear-
cut d i f f e r e n t i a t i o n o f
c o u n t r i e s .
In response to α one standard deviation shock in the European CPI level, oil Core coun
tries, except for Belgium, disploy α maximum response within six months. While
Germany has α permanent negative response to the shock, Denmork shows just the
opposite, namely α positive long-run impact of the shock to the European CPI level. In
the cose of the other countries, however, we ore unable to identify α common response.
In terms of magnitude and duration, the Italian response to the shock could put it in the
Core group.
Results for Core and Periphery industrial production IRF's ore not presented here, as they
show no marked differences between countries belonging to either group.
Overall, monetary shocks may not hove similar effects in European countries. Exomples for
this are Denmark, which shows α much larger response than the other EU countries;
France, where the maximum effect occurs offer that for Germany; Greece with α rather
errofic IRF; Portugol, where the initial effect is negoHve and never really turns posifive; and
Spain, for which the effect is lorge and extremely long-lived. The other nominal shock,
namely α one standord deviofion long-term interest shock, leods to ο more clear<ut differ-
enfiafion of countries on the basis of the transmission mechanism. With the possible excep
fion of France, all Core countries have similar IRF's in terms of sizes and durofion, which
is also true for the UK. The remaining Europeon countries, on the other hand, do not seem
to shore the transmission mechanisms of Germany and its smaller neighbours. With sev
eral notable excepfions, the division of the EU into α Core ond Periphery on the basis of
the nominal evidence presented here seems to be jusfihed. However, while France and
Denmark do not shore oil the similorifies with Germany, the UK does, putting the division
of Bayoumi and Eichengreen (1992a,b) into doubt, at least on the nominal side.
Turning to the incidence and propogafion of real shocks, we hnd that among Periphery
countries, Italy's IRF is closer to the typical response of Germany and its neighbours than
any of the other Periphery ones. The UK, for example, has on IRF of similar durofion to
Core countries, but the initial impact of α CPI shock is much higher. In general, though,
shocks to Periphery countries ore longer-lived than for Core countries. This distinction
disappears when looking at the propagation of industriol production shocks, for which
there is little difference between Core and Periphery countries. As such, Ballabriga et al.'s
(1993, 1995) result of important asymmetries with respect to infiation ore conhrmed.
8. S u m m a r y a n d Conclus ions
From the estimation results, we con clearly see differences between the tronsmission
channels for the member states of the European Union. This result emerges not only from
the presentofion of the variance decompositions in Table A, but also from the IRF's in
68 EIB Pop-ys Volumel Noi 1996
Figures 2 through 4, which give α demonstration of how different convergent and closely
linked European economies can be.
In general, our invesfigotion of the degree of symmetry of external shocks has focused
on two conditions, namely the common origin of shocks and their symmetric transmission
pattern in terms of sizes and persistence. Of the four common shocks considered, three
could be regarded os being osymmetric on the basis of the variance decomposifions, as
they emerge os α relevant source of variability of the domesfic variable for subsets of
countries only. These ore shocks to European interest rotes, the European CPI level and
European industrial producfion. The extension of the analysis to include the IRF's gen
erally conhrmed these results, in the sense that the responses of EU economies to these
shocks ore asymmetric. However, it is difhcult to hnd α large overlap between the sub
sets of countries from the variance decompositions ond those from the IRF's.
O f t h e f o u r c o m m o n
s h o c k s c o n s i d e r e d ,
Hiree c o u l d b e r e g a r d e d
a s b e i n g a s y m m e t r i c
o n the b a s i s o f t h e
v a r i a n c e d e c o m p o s i
t i o n s . These a r e s h o c k s
t o E u r o p e a n i n t e r e s t
r a t e s , t h e CPI, a n d
i n d u s t r i a l p r o d u c t i o n .
The IRFs g e n e r a l l y c o n
f i r m e d these resul ts .
Our invesfigotion into the justificofion of grouping European countries into α Core and
0 Periphery has come up with the result that it may not necessarily be possible to asso
ciate EU countries with the groupings identified by Bayoumi and Eichengreen (1992a,b).
As such, common shocks to aggregate European variables con hove very asymmetric
effects omong European economies. With important exceptions, responses of Core
countries ore very similar, no doubt refiecting similar transmission mechanisms.
However, according to the analysis, the Core should only be composed of Belgium,
Germany and the Netherlands, as the monetary evidence for Denmark and France is
different from that of Germany ond its neighbours. The posifion of the UK in the
Periphery is also less clear-cut, os its propagation of monetary shocks is very similar to
that of Germany. Long-term European interest rotes ore - on the basis of the partial cor
relofion coefhcients - important for all countries considered. The evidence on the reol
side is different in the sense that both European real variables have α strong infiuence
on most of their notional counterparts. This is above all true for α symmetric European
infiofion shock, which con hove very asymmetric effects. Instead of concentrating on the
nominal side alone, any further analysis of the benehts and costs of monetary union by
prospective entronts moy thus have to take the "realities" of the real side into account.
Consequently, on "out" wishing to join EMU - and rushing to meet the Maastricht cri
teria in order to do so - should be aware of the asymmetries on the real side and the
associated implicofions of having to follow policies designed for the Core. The success
ful membership of EMU may thus hove to include structural reforms - of the labour mar
kets, for example - to bring the transmission mechanisms of the domestic economy in
line with that of the other EMU members. Leaving aside the ubiquitous threat of the Lucos-
critique, "outs" could thus find that by rushing into EMU and ignoring those structural
measures, policies suited for α Core could entail signihcont costs, even if the process of
joining EMU as soon os possible was desirable in and of itself.
EiB Pür)i"s
Volumel Not 1996 69
A p p e n d i x A : Va r i ance Decompos i t i ons
Given below ore the forecast error vorionce decompositions of the responses of the four
dependent notional variables to innovafions at both the European and the notionol level.
Table A
Variance decomposifions: Percentages of domestic variables explained by European
and Domestic Shocks
Belgian System: Innovation to
EUM
1.9
0.5
0.4
2.7
EUM
1.6
0.6
0.1
1.0
EUM
2.4
2.2
5.4
5.1
EULR
3.7
28.5
7 1
1.3
EULR
15.6
15.4
3.6
2.7
EULR
2.7
59 .2
13.6
3.0
EUCPI
0.3
3.0
21.8
1.3
EUCPI
0.9
5.8
5.4
3.4
EUCPI
1.6
3.0
15.6
3.4
EUlP BGM BGLR
Dependent Variable: BGM
1.8 88.7 1.6
Dependent Variable: BGLR 11.9 5.4 45.6
Dependent Variable: BGCPI 8.9 6.3 4.9
Dependent Variable: BGIP 26.0 4.7 0.4
Danish System: Innovation to
EUlP DKM DKLR
Dependent Variable: DKM 2.7 73.8 1.3
Dependent Variable: DKGLR 1.9 4.5 69.7
Dependent Variable: DKCPI 1.8 1.9 1.1
Dependent Variable: DKIP 8.3 3.9 1.9
French System: Innovation to
EUlP FRM FRLR
Dependent Variable: FRM 4.8 74.0 0.7
Dependent Variable: FRLR 5.6 0.3 26.0
Dependent Variable: FRCPI 5.0 0.1 0.8
Dependent Variable: FRIP 37.2 0.7 0.7
BGCPI
2.0
4.9
49.9
0.4
DKCPI
4.1
4.8
86.1
1.2
FRCPI
13.6
3.3
59.2
6.9
BGIP
0.1
0.1
0.0
63.2
DKIP
0.1
0.3
0.1
7 7 1
FRIP
0.1
0.5
0.5
43 .0
70
EIB Pape-s
Volumel N o i 1996
German Sysiem: Innovation to
EUM
0.4
1.6
2.0
4.8
EUM
2.8
3.8
5.6
1.6
EUM
21 .7
3.3
16.6
4.2
EUM
0.4
1.1
2.6
0.3
EULR
6.4
60.1
34.8
10.7
EULR
4.5
9.8
3.5
2.6
EULR
3.5
25 .0
12.0
6.6
EULR
4.5
41 .8
22.2
3.8
EUCPI
3.8
0.5
12.3
0.3
EUCPI
2.7
2.9
5.0
3.9
EUCPI
1.1
3.0
5.5
2.3
EUCPI
2.2
1.5
10.7
2.3
EUlP BDM BDLR
Dependent Variable: BDM
4.1 74.5 3.3
Dependent Variable: BDLR
1.2 2.2 20.1
Dependent Variable: BDCPI
0.3 3.2 1.4
Dependent Variable: BDIP
25.6 0.9 1.9
Greek System: Innovation to
EUlP GRM GRLR
Dependent Variable: GRM
7.8 76.3 2.5
Dependent Variable: GRLR
5.0 1.7 73 .0
Dependent Variable: GRCPI
7.5 2.0 12.3
Dependent Variable: GRIP
8.2 0.5 5.9
Italian System: Innovation to
EUlP ITM ITLR
Dependent Variable: ITM
2.3 63 .4 2.8
Dependent Variable: ITLR
5.8 1.7 60.3
Dependent Variable: ITCPI
6.6 0.3 1.0
Dependent Variable: ITIP
30.4 1.3 1.4
Dutch System: Innovation to
EUlP NLM NLLR
Dependent Variable: NLM
1.0 81.3 5.9
Dependent Variable: NLLR
1.6 12.5 29.1
Dependent Variable: NLCPI
1.4 10.5 1.9
Dependent Variable: NLIP
7 1 8.0 3.0
BDCPI
7.4
14.2
45.8
8.5
GRCPI
2.9
3.1
63.5
2.0
ITCPI
5.3
0.9
57.8
2.9
NLCPI
2.7
9.3
48 .4
2.2
BDIP
0.1
0.3
0.3
47.1
GRIP
0.5
0.7
0.7
75.3
ITIP
0.1
0.1
0.2
51 .0
NLIP
2.0
3.2
2.3
72 .6
Volumel Noi 1996 71
Portuguese System: Innovation to
EUM
1.9
8.0
8.4
4.1
EUM
0.1
0.2
0.1
0.5
EUM
7.5
4.0
10.4
4.3
EULR
10.7
6.2
4.7
3.7
EULR
2.6
11.8
4.2
2.6
EULR
1.7
20.6
6.0
1.2
EUCPI
2.1
0.9
14.8
0.6
EUCPI
1.5
2.9
25.2
1.8
EUCPI
2.4
0.5
22.5
0.6
EUlP PTM PTLR
Dependent Variable: PTM
1.9 7 7 1 3.2
Dependent Variable: PTLR 4.5 10.9 65.7
Dependent Variable: PTCPI 8.3 8.4 0.9
Dependent Variable: PTIP 6.9 12.3 1.1
Spanish System: Innovation to
EUlP ESM ESLR
Dependent Variable: ESM 16.7 74.6 1.0
Dependent Variable: ESLR 11.3 10.5 61.2
Dependent Variable: ESCPI 16.5 12.2 1.0
Dependent Variable: ESIP 24.3 9.3 1.4
UK System: Innovation to
EUlP UKM UKLR
Dependent Variable: UKM 1.6 79.3 1.4
Dependent Variable: UKLR 2.1 13.0 56.8
Dependent Variable: UKCPI 5.4 16.2 1.2
Dependent Variable: UKIP 14.1 3.9 2.0
PTCPI
2.0
1.7
53.1
3.0
ESCPI
3.1
1.4
39.9
1.7
UKCPI
1.8
0.5
32.0
1.9
PTIP
0.6
2.2
1.5
68.3
ESIP
0.4
0.7
0.9
58 .4
UKIP
4.3
2.6
6.4
72.1
Countries are arranged in alphabetical order. Shown are the percentage decomposi
fions of the variance of the forecast error for each variable. The value given is the over
age of the one- and 60-month ahead forecast error variance decomposifion. The notes
to Table 2 give on explanation of the variables.
72
; 'B ^accis
Volumel Noi 1996
A p p e n d i x Β: A n I n t r o d u c t i o n t o B a y e s i a n VAR M o d e l l i n g ( 16)
The point of departure for the Boyesion VAR approach is the standard VAR model used
to derive short-run linkages between economic variables of interest. With long log
lengths and mony variables, however, the number of observations available will often
not be enough to allow α reliable estimofion of all the coefficients in the VAR. In order
to circumvent the likely problem of over-parometerisotion, we hove to take recourse to α
Boyesion estimofion method of the VAR, which combines prior beliefs with trodifionol
VAR estimation methods.
The basis is provided by Boyes' theorem, from which it follows that the conditional prob
ability of the coefhcient vector (π) to be estimated given the observations (X), ρ(π|Χ),
is α function of the probability of our prior beliefs about the parameters, ρ(π), the condi
tional probability of the observations given the coefhcients, p(X | π), and the probabil
ity of the observations, p(X), i.e.
[B.l] ρ(π|Χ1 = ρ(π)ρ(Χ|π)/ρ(Χ)
This relofionship underpins Boyesion inference and lies at the heart of the esfimation pro
cedure, where sample information is combined with prior distributions on the coefficients
to give final estimates.
An Application of Bayesian VAR Analysis: The Minnesota Prior (17)
The central role of our prior beliefs about the coefhcients in the esfimation procedure con
be easily seen from the inclusion of ρ(π) in [ B . l ] . In procfice, the convenfion of the
Minnesota prior provides α sensible set of Boyesion prior beliefs that hove become the
standard in Boyesion VAR models. Under this assumpfion, coefficients are set in accord
ance with the random walk hypothesis - or α rondom walk with drift - for the variables.
This is implemented by using α mean of zero for the prior on oil coefficients except the
hrst own log in each equation, even though the data is allowed to override this restric
tion in the actual esfimation process.
Specification of the Priors
In order to reflect the inherent uncertainty of the coefficients, the distribution of the prior
is mode to depend upon α small parameter vector, τ, that controls aspects of the prior
for which we lock knowledge, such os its mean. Estimofion of the model in fact proceeds
by choosing τ.
/ 6} Approachable introductions to Bayesian VAR modelling in the literature are Todd ( 1984), Litterman ( ! 984,
1986} and Runkle (1987).
17) The Minnesota prior takes its name from the fact that it was developed by economists at the University of
Minnesota and the Federal Reserve Bank of Minneapolis in Minnesota.
E :B " α ο ί 'S
Volumel Noi 1996 73
Finding the Optimal Prior
Two opfions ore open to the researcher for finding the optimal prior. She can either chose
0 standard prior, which is associated with α specihc setting of τ that hos worked well in
the post or refiects some empirical rule-of-thumb concerning time-series behaviour, or she
con strive to hnd on optimal prior. A reasonable criterion for the latter is to select the prior
associated with the set of parameters, τ * , that maximises the likelihood funcfion of the
model. Table Β presents the results of such α procedure for finding the opfimol prior. The
four possible prior specihcofions were no prior and α loose, medium and tight prior re
spectively. (18) With the four specifications, α procedure was used that minimises the
logarithm of the determinant of the covoriance matrix of the system's forecast errors.
Table Β
Log determinants of the out-of-somple forecast error covoriance matrix
System
Belgium
France
Germany
Greece
Italy
Netherlands
Portugal
Spain
UK
Unrestricted
-81.35
•87.44
•90.31
-78.06
•80.47
-83.46
-78.29
•78.95
-86.96
Prior
Loose
-84.06
-90.50
-92.18
-80.80
-86.72
-87.44
•82.12
-81.96
•90.52
Medium
•86.66
-93.55
•94.43
-83.53
-90.46
-92.38
•86.72
-86.10
-93.97
Tight
-86.84
-94.15
-94.73
-82.35
•91.56
-92.10
•86.75
•87.92
•95.18
Minimum logarithms of the determinant of the estimated vorioncecovorionce matrix ore
denoted in boldface.
We con see that with the exception of Greece and the Netherlands, the tight prior per
forms best out-of-somple in the sense that it minimises the logarithm of the determinant
of the estimated vorionce-covorionce matrix. For the other two countries, the medium
prior gave the desired outcome. Subsequently, the tight prior was imposed on Greece
and the Netherlands as well os Denmark, for which the opfimol prior could not be cal
culated directly. (19)
/ 8} The coefficients are assumed to be normally distributed, a n d thus completely described by their means and
standard deviations. The mean of the distribution is one on the first lag a n d zero on al l lags greater than one,
while the standard deviation is determined by a number of factors, including the overall tightness of the prior
around the mean a n d the tightness of higher lags relative to the hrst lag. The latter two parameters have been
set to range from loose to tight in the selection procedure.
19} This method of determining a prior was not possible in the case of Denmark, where monthly industrial pro
duction hgures were not available for the year 1994.
EIB PoDors
7 4 Volumel N o i 1 9 9 6
Estimating the Bayesian VAR
After the selecfion of the opfimal prior, estimofion proceeds vio the Theil-Goldberger
mixed-estimafion procedure - os described in Theil (1971) - of α system of two sets of
equotions, one for the actual doto and one for the prior. The two equafions are then esfi
mated simultaneously by the application of feasible generalised least-squares.
Ì:B Pape-s Volumel Noi 1996 75
References:
Bollobrigo, E, Sebastian, M. and Voilés, J. (1993). "Interdependence of EC economies:
0 VAR approach." Banco de Espaiïa Servicio de Estudios Documento
de Trabajo η.- 9314. Madrid: Banco de Espono.
Ballabriga, E, Sebastian, M. and Voilés, J. (1995). "European asymmetries."
ESADE Papers No. 141. Madrid: Escuelo Superior de Administrociòn y
Dirección de Empresos, June.
Bayoumi, T. and Eichengreen, B. (1992α). "Shocking ospects of European monetary
unihcofion." CEPR Discussion Paper No. 643, May.
Bayoumi, T. and Eichengreen, B. (1992b). "Is there α confiict between EC
enlargement and European monetary unification?"
CEPR Discussion Paper No. 646, May.
Bofinger, Ρ (1994). "Is Europe on optimum currency area?" In 3 0 Years of European
Monetary Integration from the Werner Plan to EMU, pp. 38-56
(ed. A. Steinherr). London: Longman.
Commission of the European Communifies (1990). "One market, one money."
European Economy No. 44, October.
Demertzis, M., Hughes Hallett, A. J. and Rummel, Ο. J. (1996). "Is α 2-speed system
in Europe the onswer to the confiict between the German and Anglo-
Saxon models of monetary control?" CEPR Discussion Paper No. 1 4 8 1 ,
(forthcoming).
Kenen, Ρ (1969). "The theory of optimum currency areas: on eclecfic view."
In Monetary Problems in the International Economy, pp. 41-59
(ed. R. Mundell and A. Swobodo). Chicago: University of Chicago Press.
Lostropes, W. D. and Koray, E (1990). "Internotionol transmission of aggregate
shocks under fixed and flexible exchange rote regimes: United
Kingdom, France, and Germany, 1959 to 1985."
Journal of International Money and Finance, 9, pp. 402-423.
Litterman, R. B. (1984). "Forecasting and policy analysis with Boyesion vector
outoregression models." Federal Reserve Bank of Minneapolis Quarterly
Review, Fall, pp. 30-41.
Litterman, R. B. (1986). "Forecasting with Boyesion vector outoregressions - five years
of experience." Journal of Business & Economic Statistics, 4, pp. 25-38.
EIB Popo's
76 Volumel Noi 1996
Masson, Ρ R. and Taylor, M. R (1993). "Currency unions: α survey of the issues."
In Policy Issues in the Operation of Currency Unions, pp. 3-51
(ed. Ρ R. Masson and M. Ρ Taylor). Cambridge: Cambridge University Press.
McKinnon, R. I. (1963). "Opfimum currency oreos." American Economic Review,
53, pp. 717-725.
Mundell, R. A. (1961). "The theory of opfimum currency areas."
American Economic Review, 5 1 , pp. 657-665.
Runkle, D. E. (1987). "Vector outoregression and reality."
Journal of Business & Economic Statistics, 5, pp. 437-454.
Sims, C. A. (1980). "Macroeconomics and reality." Econometrica, 48, pp. 1-48.
Tovlos, G. S. (1993). "The "new" theory of opfimum currency areas."
The World Economy, 16, pp. 663-685.
Theil, H. (1971). Principles of Econometrics. New York: John Wiley & Sons.
Thygesen, N. (1995). "The prospects for EMU by 1999 - and refiecfions on
arrangements for outsiders." Lecture for the European Policy Research
Unit, Manchester.
Todd, R. M. (1984). "Improving economic forecasting with Boyesion vector
outoregression." Federal Reserve Bank of Minneapolis Quarterly
Review, Foil, pp. 1 8-29. Reprinted in Modelling Economic Series,
pp. 214-234 (ed. C. W . J . Granger, 1990). Oxford: Clarendon Press.
Von Hogen, J. (1993). "Monetary union and hscal union: α perspective from fiscal
federalism." In Policy Issues in the Operation of Currency Unions, pp.
264-296 (ed. Ρ R. Masson and M. Ρ Taylor). Cambridge: Cambridge
University Press.
f :B Par:)i"s
Volumel Noi 1996 77
Is Maastricht α g o o d contract?
Bernhard Winkler
European University Institute, Florence. EIB Campil l i-Formentini Scholarship 1994-1996
1 . The M a a s t r i c h t t r a n s a c t i o n
It is to state the obvious to soy that the treaty of Maastricht is α contract. Surprisingly,
there has not yet been any economic analysis of the obvious. This paper proposes to
consider the Maastricht Treaty os α contract device that must organize α mutually bene
hciol tronsocfion, here European Monetary Union (EMU). In order to assess whether
Maastricht is α good contract three possible functions of contracts ore explored. First,
the treaty con provide countries with extra incenfives for desirable behaviour. Second,
it can co-ordinate behaviour among different countries to produce α desired outcome.
Third, the treaty con structure behaviour and decision procedures such that relevant
information is revealed. As for the content of the Maastricht Treaty, three essential fea
tures con be distinguished. Maastricht aims to establish α single currency for Europe, it
seeks to ensure ο stable currency and hnolly it mokes entry to EMU conditional on ful
hlling convergence criteria and mandatory for the countries sotisfying the criteria (with
the excepfion of the UK and Danish opt-outs).
The k e y t o u n d e r
s t a n d i n g the M a a s t r i c h t
c o n t r a c t a n d the r o l e
o f t h e c o n v e r g e n c e
c r i t e r i a is t o r e c o g n i z e
t h a t interests a n d p r i
or i t ies d i f f e r across
countr ies a n d c h a n g e
o v e r t i m e .
As explained in the "strategic view on EMU" proposed in Winkler (1996a) the key to
understanding the Maastricht contract and the role of the convergence criteria is to recog
nize thot interests and priorifies differ across countries and change over time. The coun
try that has most to lose from EMU both in terms of credibility as well os in terms of sacri
ficing sovereignty is Germany. Therefore, at Maastricht Germany was willing to trade
away the Bundesbank and share sovereignty equally only if the single currency was to
become "at least os stable os the deutsche mark". Thus there was little question that the
statutes of the European Central Bank (ECB) would hove to be modelled closely offer the
Bundesbonk's. However, central bonk independence can at most be seen os α necess
ary condition for enduring price stability: it is certainly not sufficient (1). This is where
the convergence criterio come into ploy. They coll for infiofion and interest rotes to be
within 1.5 percentage points of the three best performers and for membership of the
exchange rote mechanism (ERM) for at least two years without devaluofion on own ini-
tiofive. The hscal criteria stipulate α dehcit of at most 3% and α public debt of at most
60% in relation to the candidate's GDP. These reference values ore to be applied to deci
de which countries qualify for entry into EMU at the end of stoge two, but the fiscal
conditions are to hold also throughout stoge three of the Maastricht process.
This paper draws on research for my PhD dissertation at the European University Institute, which was funded by the
European investment Bank. Helpful comments by Michael Artis and Spyros Vassilakis are gratefully acknowledged.
1) The extensive literature on the credibility of monetary pol icy points to contractual solutions in the form of
rules (e.g. on money supply), delegation to a conservative central bank (Rogoff, 1985) or the use of explicit
incentive contracts for central bankers as ways to achieve low inflation (Walsh, 1995). Alternatively, under
repeated interaction, incentives to bui ld and preserve a g o o d reputation can also lead to low inflation (Backus
a n d Driffill, 1985). Ultimately, however, price stability requires the consensus a n d support of the population at
large over and above the institutional details of monetary policy making.
': iB PoDc's
Volumel N o t 1 9 9 6 7 9
The Maastricht criteria hove drown severe criticism from economists, who tend to con
sider them OS arbitrary non-sense (Buiter et a i , 1993) or os either superfluous or, worse,
OS harmful and self-defeating (De Grauwe, 1994). They seem to hove little to do with
economics. In poriiculor they have little to do with the requirements for on opfimum cur
rency area, such os price and wage fiexibility, factor mobility or fiscal transfers. Instead,
oil the criteria ore best interpreted os indicators of (post, present and future) credibility.
In short, in stage two candidate countries ore asked to demonstrate their stability orien-
tafion before joining EMU. The reasoning is that only α deeply rooted "stability culture"
among EMU members will allow the ECB to produce stable prices at low real costs. In
particular, conflicts between the orientation of fiscal and monetary policy ore to be avoided.
In order to make sense os on entry condition for EMU the convergence criteria must satis
fy two requirements. The behaviour they induce in stage two must have lasting ond bene
hciol effects on stage three. Second, they must be seen to address inefficiencies from
spillovers, i.e. induce desirable behaviour that would otherwise not be in the individual
self-interest. The criteria ore important if producing stability, credibility and reputation for
EMU has public good features, i.e. if it requires individual sacrifice for the common good.
This secfion has characterized the Moastricht transaction os on exercise in the pooling
of sovereignty in α single currency, α "selling" or "sharing" of the Bundesbank, in return
for the acceptance of Germon standards for α stable currency. Statutory independence
of the ECB was regarded as insufficient and hod to be supplemented by ο mechanism
to organize the producfion of credibility and reputation via the convergence criteria.
Section 2 explores possible rationales for the Maastricht Treaty provisions by asking
what would happen in the absence of any contractual devices. Indeed some commen
tators such OS De Grauwe (1993) hove suggested that the decision to form/join EMU
should be entirely voluntory and unconditional. It is then invesfigoted how contracts may
help in organizing EMU. In Section 3, the principal funcfions of the Maastricht entry
conditions are analyzed. Section 4 concludes with policy implications.
2. Maastricht games a n d contracts
The Maastricht game
The strategic view on EMU starts from the premise that costs and benehts differ across
countries and in their time prohles. In porficulor countries with α high domestic monet
ary credibility ore concerned about α possible loss of reputofion and price stability in
EMU. We coll the advocate of these interests "the Principal" for the rest of the paper,
most obviously represented by Germany and the Bundesbank. It prefers that conver
gence and credibility be established by national effort prior to admission into EMU. For
F Β ί'αρ ;·5
8 0 Volumel N o i 1 9 9 6
low credibility countries, on the contrary, the whole point of EMU is to gain credibility
more cheaply, so they prefer convergence inside EMU, if at oil. This group includes most
candidate countries and will be called "the Agent" throughout the paper (2). In order to
concentrate on the strategic interaction between these two groups of countries consider
the following objective functions for Principal (P) and Agent (A) respectively.
In s h o r t , c a n d i d a t e
c o u n t r i e s a r e a s k e d t o
d e m o n s t r a t e t h e i r
s t a b i l i t y o r i e n t a t i o n
b e f o r e j o i n i n g EMU.
(1)
(2)
V(P) =ρ(Τ,+ 0ωΕ) + (1 -0)ωΕ
U ( A ) = p T , - ^ e
The first term in both equafions captures fhe total expected net benefits from EMU, where
p is the probability and timing of EMU. T, and T,, ore the (net) benefits of stage three of
EMU for Principal and Agent respectively. It is reasonable to assume that T^ is negofive,
i.e. the Principal would not agree to EMU in the absence of any convergence (E). E only
concerns the externality component of convergence, i.e. it abstracts from the conver
gence effort that α country would find in its own interest to undertake in preparation for
EMU. For the Agent the extra Maastricht-induced component of convergence is costly
with increasing marginal costs. The higher β the more painful it is for α country to pur
sue rigid hscol and monetary policies or unpopular reforms in preparation for EMU.
The Principal, on the other hand, is interested in inducing as much prior convergence as
possible, where w is his morginol utility of convergence. There ore two possibilities: the
Principal might be interested in convergence per se or he cores about it only if EMU hap
pens. In on alternative interpretation, fraction 0 of convergence is reversible and thus
will be lost if EMU does not materialize. The shore (I-0) refiects durable convergence,
independent of EMU, or the temporary ufility that even reversible convergence yields
during the fime it is forthcoming. In the special cose of 0=0, the degree of convergence
in stage two has no particular value for stage three. For 0=1 convergence only matters
for the Principal in stage three and only if EMU comes about.
Imagine α simultaneous move gome with objective funcfions (1) and (2) where the
Principal must decide whether to surrender the Bundesbank for EMU (p=l) or not {p=0)
and the Agent decides on the amount E of convergence to undertake. For illustration the
following numerical values ore assumed henceforth: Tp=-1, Τ, =4, (o=2 and β=2. In the
payoff matrix (Figure 1) the Agent's payoff for each combination of strategies is given
first, the Principal's is the second term for each outcome.
2} For a more detailed account of the role of the criteria in the Maastricht negotiations see Bini-Smaghi et al.
(1994).
Volumel Noi 1996 81
Figure 1. The Maastricht gome (Nash)
Principal
Agent Opfimal Convergence: £= (=(dß) No Convergence: E=0
EMU (ρ=λ)
1, 1
2,-1
No EMU (p=0)
- 1 , (1-0)2
0 , 0
The unique Nosh equilibrium is in the bottom-right corner of Figure J: no convergence
is forthcoming and EMU does not happen. The co-operofive solution that maximizes joint
welfare colls for EMU to happen and for the optimal convergence effori which balances
the marginal cost of convergence to the Agent with its marginal beneht to the Principal.
The co-operative solufion in the top-left corner, os in the well-known Prisoners' dilemma,
is not sustainable since, once EMU is assured, the Agent has no incenfive to undertake
costly convergence. Given that, the Principal will not agree to EMU. Conversely, if the
Agent provided optimal convergence the Principal will still refuse EMU unless ο > l / 2 ,
i.e. unless there is enough EMU-specihc convergence that the Principal con only secure
by gronfing EMU. An example of ο big 0 would be the fear that the single market,
exchange rate stability and the entire convergence process could unravel unless it is
"locked-in" via EMU. A small 0 would obtain if countries' convergence behaviour in
stage two said nothing about their reliability for stage three or, on the contrary, conver
gence would continue just the some even in the absence of EMU.
The c o - o p e r a t i v e
s o l u t i o n , a s i n t h e
w e l l - k n o v i f n P r i s o n e r s '
d i l e m m a , is n o t sus
t a i n a b l e .
For concreteness, coll the players Germany and Italy. Germany holds the key to EMU
coming about; Italy con choose convergence (soy hscol rectitude) or otherwise. If
Germany commits to EMU it must fear that ex post, with the Bundesbank surrendered,
Italy will not produce sufhcient and durable convergence. Italy may not resist the temp-
tofion to try to hove Europe boil out its debt, redirect its priorities towards employment
instead of price stability, delay hscol reform further etc. Conversely, Italy may fear that
painful adjustment E would not be rewarded with EMU entry.
Contracts
There ore severol ways, in principle, in which the Maastricht Treaty may improve on the
inefhcient Nash equilibrium of Figure 1. First, by structuring the gome by specifying α
move order, i.e. when decisions are token. Second, by allocofing decision authority, i.e.
who decides what. Third, by altering the payoffs of the gome, e.g. by committing
players to certain actions, outcomes or procedures, where breach of treaty carries α
penalty. In particular, the treaty con specify decision rules, i.e. regulate on what basis
and how decisions ore token. Here, the Maastricht criteria examined in Section 3 ore α
82
E IB E'opc-s
Volumel Not 1996
prime example. The Maastricht Treaty deploys α combination of all three options, which
will be explored in turn.
The m o s t i m m e d i a t e
a n s w e r w o u l d b e t o
p o o l a l l a u t h o r i t y a t
t h e E u r o p e a n l e v e l .
The first simple measure the treaty con take is to prescribe α particular move order, i.e.
hove the players in Figure I make their choices sequentially. Then if the Agent moves
hrst, he will choose the minimum convergence E that is necessary to just entice the
Principal to go along with EMU. If T,. is negative, os before, all that is needed is α posi
tive 0, i.e. that some of the prior convergence is EMU-specihc. Assume 0 = 3 / 4 for illus
tration; then the top-left corner in Figure 2 will become α Stockelberg equilibrium, i.e. α
Nosh equilibrium in α gome where the Agent moves first or con pre<ommit to this pre
ferred acfion (3).
Figure 2. The Maastricht gome (Stockelberg)
Agent
Principal
Minimal Convergence: E=2/3i'=-T,,/0(yj
No Convergence: £=0
EMU (p=])
3 1/3, 1/3
4,-1
No EMU (p=0)
-2/3, 1/3
0,0
Note that the reverse move order, where the Principal commits first, is of no help. In this
cose the Agent would always respond with zero convergence and hence the Principal
would refuse EMU. Note also that the convergence £ induced by the efficient move
order, will not in general correspond to the efficient amount of convergence of the co-ope
rative solution in Figure I. In the above example the minimal effort is subopfimol (i.e. 2/3
rather than one), but for 0 < l / 2 convergence becomes excessive rather than deficient.
This inefficiency may be one reason why the Maastricht Treoty not only colls for "conver
gence first" but also sets minimum convergence requirements. Moreover, Maastricht esta
blished α final deadline for EMU in 1999, together with the criteria, and therefore does
not leave the decision on EMU in the Principal's hands. This suggests that the move order
alone was perceived os insufficient to guarantee on efficient transition to EMU.
The second possibility to improve on the Nosh outcome of the Maastricht gome
concerns the allocation of decision authority. The most immediate answer to the
Prisoners' Dilemma of EMU would be to pool all authority at the European level. If
Europe hod already ochieved full political union, joint decisions would refiect European
welfare (or the result of intergovernmental bargaining) and could be legitimately exe
cuted and enforced even ogoinst individual nations' interests. For now, however, it
seems reasonable to ossume that contracts and explicit treaty commitment are necessa
ry for those purposes. Then α two stage game con be envisaged, where, hrst, players
3} Note thot the bottom-right corner still is a Nosh equilibrium, but it is not "subgame perfect": If fhe Agent can
move hrst, he wil l choose to converge, since he knows that the Principal's best response wil l then be to al low
EMU to g o through. Strictly speaking E needs to be slightly greater than 2 / 3 in order to break the tie.
EiB Parx?.'s
Volumel N o i 1 9 9 6 83
contract over decision rights and, second, play α Nosh gome in the decision variables
allocated previously. If it were possible to contract for α "reverse assignment" of deci
sion rights, then the Agent would decide whether EMU would go ahead and the
Principal could choose the degree of convergence os in Figure 3.
Figure 3. The Maastricht gome (reverse assignment)
Principal
Agent EMU(p=l):
No EMU {p=0]
Maximal Convergence:
£=2 (=V2T,/i3)
0,3
-4,(1-0)4
E=0
4,-1
0 , 0
The M a a s t r i c h t c r i t e r i a
a t t e m p t t o i n t e r n a l i z e
t h e e x t e r n a l i t i e s i n t h e
o r i g i n a l M a a s t r i c h t
g a m e a n d t r y t o
a c h i e v e t h e co-oper
a t i v e s o l u t i o n .
Here the unique Nosh equilibrium is the outcome in the top-left corner. The Agent will
olwoys wont EMU to happen and the Principal wonts to extract the maximal conver
gence, £ , which will leave the Agent just no worse-off than in the absence of Maastricht.
The treaty contoins some elements of α reverse assignment. At least on paper, it assures
that p= 1, i.e. that EMU will happen for sure by 1999 at the latest and it prescribes (qual
ified) majority vote for the entry decisions. This means the Principal could be outvoted:
he cannot block EMU single-handedly. As for convergence, the Principal has been allow
ed to impose the Maastricht criteria and also to ploy α vocal role in their interpretation.
Moreover, since several of the criteria ore formulated in relafive terms, by setting mon
etary policy for the DM-block of currencies the Bundesbank effecfively determines the
absolute values of the infiofion and interest rote criteria.
While the reverse assignment allows α superior outcome compared to the original "natu
ral assignment" in Figure 1 it runs the risk that the Principal imposes excessive conver
gence on the Agent, as compored to the co-operofive solution (in the example, the maxi
mum £ = 2 instead of £=1). Moreover, once the treaty is concluded and if the Principal
con effecfively control the Agent's effort, he could always ask for sfili more convergence
and could moke the Agent worse off than without Moastricht. Anticipating this the Agent
would refuse to sign away his control over convergence at Maastricht. The main pro
blem with the reverse assignment is that it is difficult to enforce, because it is "unnatu
ral". Certainly the Bundesbank hos been already signed owoy at Maastricht and
Germany con be outvoted in the Council, but still it would be hard to conceive that it
could be really coerced into EMU against its will in 1999. Likewise, Germany certainly
cannot dictate convergence policies of sovereign partner countries, even with the most
rigid interpretafion of the criteria (4).
4} The Principal can impose addit ional convergence by exploiting the fact that for any given level of conver
gence ex post, the Agent prefers EMU to N o EMU. Further risks to Ihe stability of the equilibrium in Figure 3
arise if the reverse assignment is not fully credible. Then, for 0 1 / 4 the Principal prefers the bottom-left outcome
and wil l try to prevent EMU, aher convergence has materialized. Similarly, the Agent hos every incentive to
cheat on convergence to try to achieve his preferred outcome (top-right).
84
E'B Pap3's
Volumel N o t 1 9 9 6
The Maastricht criteria can be seen os α way around the problems of the reverse assi
gnment. First they seek to protect the Agent against demands for excessive ond ever
greater convergence. Second, they reossure the Principal by making his commitment to
EMU condifional on sufficient prior convergence. Third and most importantly, by making
entry to EMU condifional on convergence it becomes in the Agent's own interest to
undertoke convergence effort, without signing owoy notional control. In this woy the
Maastricht criteria attempt to internalize the externalities in the original Maastricht gome
and try to achieve the co-operative solution in Figure 1.
The M a a s t r i c h t c r i t e r i a
h a d t o b e s i m p l e ,
v is ib le a n d e q u a l f o r
a l l c a n d i d a t e countr ies.
It was shown how the Maastricht Treaty might help to solve the Prisoners' Dilemma of
the Maastricht gome by specifying who decides what and when. The obvious thing,
however, would be to commit to the desired co-operofive outcome directly. A complete
confingent contract would specify the actions to be undertaken by the two sides under
oil conceivable circumstances and would be perfectly enforceable. Real life contracts,
however, ore usually incomplete because of transaction costs (Hart and Holmstöm,
1 987). These arise, hrst, from the difficulty of anticipating oil possible eventuolifies. The
ERM crisis of 1992/93 is ο prime example of such an unforeseen contingency. Second
there are costs of agreeing and deciding. Third, the imprecision of language mokes it
difficult to give clear and unambiguous descriptions of the relevant states of the world.
Fourth, legal enforcement of contracts is often difficult ond costly. Enforcement will be
particularly hard if the subjects ore sovereign states and the union has only α limited
capacity to impose formal punishment.
A further set of problems arises under private informafion, i.e. the case where one of the
two parties has superior knowledge about the state of the world. In particular, policy
makers' true preferences and intentions might not be known to the other party and rele
vant actions might not be easily observable or at least not verihoble in court.
Convergence effort, i.e. measures that contribute to fhe credibility and stability of EMU
both in the run-up to EMU and in stage three itself, ore hordly contracfible (and there
fore enforceable) directly, even if they should be observable. For example, it is hard to
imagine a contract ruling out all conceivable manipulations of budget figures which
have nothing to do with achieving the sustainable and sound public finances the treaty
is interested in.
The Maastricht criteria, therefore, must be understood os α highly imperfect substitute for
α fully state-contingent and enforceable "ideal" complex contract. They had to be
simple, visible and equal for oil condidate countries, and of course bear at least some
relation to the underlying variable of interest, i.e. the willingness and capacity to sup
port and sustain stability oriented policies. However, the cost of abiding by α crude and
E-BPocrrs
Volumel Noi 1996 85
infiexible contract to the letter could be very high if important informotion is disregarded
or sizeable shocks (e.g. recessions) intervene in the meantime. A way to get around this
problem is not to commit to porficulor actions or outcomes directly, but to conclude "relo-
fionol contracts" (Milgrom and Roberts, 1992). These agree on the objecfives and the
criteria, the process and procedures of decision making, not on the decisions them
selves. By providing α framework for the decision to move to stage three of EMU, i.e.
by dehning who decides when, what and how, the Maastricht Treaty constitutes such α
relational contract. As such it preserves the commitment value of α contract (which is
necessary to sustain α co-operative solution) while preserving valuable fiexibility in the
light of unforeseen confingencies.
The principal elements of this relofionol contract concern the move order discussed before,
the outomoticity of EMU in 1 9 9 9 and the joint European decision making on the appli
cation of the convergence criterio. Convergence has to precede EMU becouse it is not
easily controctible and enforceoble ex post. This contrasts with the transfer of sov
ereignty from the Bundesbank to the ECB, which is highly observable and controctible.
Automoticity tries to ensure the Principal's commitment to EMU and, more generally, the
mandatory participation of high credibility countries, who would confer α positive
spillover on EMU by joining. Joint European decision making (abstracting from the
precise voting rules) means that oil parties affected by the decisions, in particular the
"outs" OS well OS the "ins", ore present and under efficient bargaining, oil externali
ties could be internalized. However, this presupposes that side-payments, e.g. on
Political Union or structural funds, are available to compensate individual countries.
However, problems arise for ex ante convergence incentives, which could suffer if countries
onficipoted renegotiation at the time of the application of the criteria. Depending on bar
gaining strength Agent countries may receive insufficient reward for prior convergence.
Since convergence costs are olreody sunk at the fime of the entry decision, the Principal may
take advantage of this and extract further convergence or impose addifional condifions.
The prospect of opportunistic behaviour will lead countries to undertake less convergence
than in the absence of renegotiafion. Thus ex post bargaining con lead to the distortion of
ex ante incentives, os in Williamson's (1975) "hold-up problem". In the presence of asym
metric information, moreover, additional bargaining inefficiencies ore prone to arise
(Myerson and Sotterthwaite 1983). Uncertainty about other countries' voluofion of EMU or
convergence costs may then prevent α mutually beneficial renegofiafion.
One possibility to limit the negative effects of incomplete contracts, that con arise if am
biguities and omissions in the treaty hove to be hlled later by biloterol bargaining, is to
provide for third party arbitration. This role is porfiy assumed by the convergence reports
to be prepored by the European Commission and the European Monetary Institute.
EIB Papeis
86 Volumel Noi 1996
While not binding, they will still be very important in providing on (ideally impartial)
interpretafion of the treaty provisions. The other olternofive is to conclude long-term
contracts, even if imperfect, and stick by them rigidly, even in the event of adverse
shocks. While this is likely to lead to inefficient convergence, at least it reduces the risk
of EMU not coming about at all. The presence of bargaining inefficiencies may explain
the dogged determinotion of European leaders to stick by "the treaty, the whole treaty
and nothing but the treaty", studiously avoiding α re-opening of the Maastricht
"Pandora's box".
O n e p o s s i b i l i t y t o l i m i t
t h e n e g a t i v e ef fects o f
i n c o m p l e t e c o n t r a c t s ,
is t o p r o v i d e f o r t h i r d
p a r t y a r b i t r a t i o n . This
r o l e is p a r t l y a s s u m e d
b y t h e c o n v e r g e n c e
r e p o r t s t o b e p r e p a r e d
b y t h e E u r o p e a n C o m
m i s s i o n a n d t h e
E u r o p e a n M o n e t a r y
I n s t i t u t e .
3 . The r o l e o f t h e M a a s t r i c h t c r i t e r i a
The previous section has already furnished several explanations for the adoption of the
Maastricht criteria by asking what would happen in the absence of any such conditions.
The task of the criteria is to organize α co-operative solution of the Maastricht gome.
Given the incomplete contract framework proposed as the appropriate reference point,
the task of the paper is not to defend (or propose) any porficulor numbers for the cri
teria. All that matters here is that policymakers cored enough about them, rightly or
wrongly, to include them in the treaty. Andreas Kees (1992), the German representofive
on the EC monetary committee, where the criteria were conceived, lists three principal
funcfions. According fo him (p.31 ), the criteria ore not of α technicol but of α political
nature. They serve as guideposts for the orientofion of economic policy, they create pres
sure for consolidation and they hove α signalling function, especially with α view to the
hnanciol markets. The moin ideo was to create α "dynomic tension", where the prospect
of α hxed deadline for stage three would induce and facilitate the necessary adjustments
much earlier and in turn create the desired momentum for EMU. The three functions of
the criteria ore discussed in the following, starting with the externol incentive argument.
Providing convergence incentives
As explained before, the Maastricht criteria can serve as α substitute for contracting for
opfimal convergence directly. For example by making the probability of EMU entry in
equation 3 depend on the degree of convergence as measured by the criteria, the in
centives of the two porties become more closely aligned. Now it is in the own interest
of the Agent countries to undertake costly convergence effort. EMU becomes the reward
that the Principal offers as α function of convergence effort. Ideally the incenfive contract
should be structured such os to completely internalize the convergence externality in the
Maastricht gome.
(3) U(A) = p ( E > - T , - ^ E '
ί iB PoriC's
Volumel Noi 1996 87
With the Maastricht criteria governing entry rather than decisions by the Principal
ond/or the Agent, the probability of entry p now is endogenous and on increasing func
tion of convergence effort £ . From maximizing equation 3 the Agent will choose opti
mal convergence effort as
(4) E^-JJ^.Ï^ dB β
Convergence effori will be higher the greater the rewards from EMU ( T A ] , the smaller
the costs of convergence (β] and the more extra effort raises the entry probability p. In
the probabilistic formulofion of equations 3 ond 4 it is already assumed thot there is
uncertainty, either about the opplicotion of the criteria or (and) about the economic
transmission mechanism which tronslotes convergence effort into the outcomes relevant
for the criteria. As shown in Winkler (1996b) the presence of uncertainty about the cri
teria con actually be beneficial for convergence incentives. The intuifion is straightfor
ward: if the criteria were totally precise countries who ore for away from fulhlment will
"throw in the towel", while countries sure about reaching them will no longer exert any
further convergence effort either. For intermediate coses convergence incentives are
weakened, rather than sharpened by uncertainty.
Thus, if the aim is to maintain α "dynamic tension" and maintain the momentum for conver
gence for the greatest possible number of member states, then ex ante some uncertainty
should be kept alive, os long as it influences policy decisions. No country should be ruled
out or ruled in, at least not publicly. This also provides α rofionole for the EMI and the
Commission not to do α "dry run" of convergence reports in 1996, at least not unfil oil the
budget measures for 1997 hove been approved. Thus the simple idea underlying equa
tion 3 provides on answer to the criticism of the criteria that argues that either they will be
useless because they will be overridden politically or that they will be harmful because they
cannot possibly be met. It is precisely the uncertainty about the flexibility and interpreta
tion of the criteria that renders the criteria useful and benehciol as an incenfive device (5).
The Waigel stability pact
As on alternative (or in addifion) to the convergence criteria, incentive effects could also
be produced by making the benehts of EMU α function of convergence. An example of
this would be the Waigel stability poet. This proposal first put forward by the German
finance minister in late 1995 colls for automatic sanctions in the form of fines for any
5} The leaked statements by German finance minister Waigel before a parliamentary committee in 1995,
where he asserted that Italy would not be in the hrst group joining EMU "and they know it", provides an ins
tructive episode. The press report caused a severe market reaction against the Lira and Italian bonds. From the
perspective of this model, this was not because the markets necessarily had a much more optimistic view on
Italian entry probabilities, but rather because common knowledge of α zero probabi l i ty would destroy Italian
convergence incentives.
E IB PoDt;rs
8 8 Volumel N o i 1 9 9 6
breach of the hscol criteria in stage three of EMU. The concern is that countries that hod
great difficulty to converge even under the threat of exclusion from EMU ore even less
likely to do so once that extra incentive has vanished. On the other hand the costs of
convergence (e.g. the parameter β in equation 3] should be lower in stage three, if inter
est rotes and inflation come down for those countries.
The s t a b i l i t y p a c t is a t
m o s t a p a r t i a l s u b
s t i t u t e f o r t h e c o n
v e r g e n c e c r i t e r i a .
In terms of our model the Waigel poet would have three main effects. First, equafions 1-3
con be applied to incentive issues in stogfe three. Then T, would be α negofive Waigel
penalty for the Agent and p(E] the probability that it will be imposed, which is now
decreasing in convergence. Countries will be disciplined from equation 4 the greater the
fines, the more the risk of incurring them depends on their behaviour and the lower the
costs of fiscal austerity. Second, the Woigel poet, therefore, would alter the parameters
of the model os applied to stage two. In particular, it should reduce the risks to the
Principal (raise Tp) or render stage two convergence more durable (increase 0). Both
should help overcome his reservofions over EMU. Third, the Waigel poet could affect the
Agent's payoffs from EMU not only via α lower T,, from the risk of incurring hnes in EMU.
More importonfiy, the benefits of EMU could become α function of prior convergence,
i.e. be written as T(Ej in equation 3. Thus they could operate in much the some way os
fhe Maastricht criteria and render it in the candidate's own interest to take correcfive fis-
col acfion before entering and thereby reduce the risk of incurring penolfies in EMU.
These incentive effects of the Waigel pact open the possibility for α trade-off, i.e. α
reloxofion of the entry conditions in return. Such α deal could moke everybody better
ofh However, os will become clear further down, α lax opplicofion of the criteria and α
large inifiol EMU will encounter less favourable storfing conditions and α lower reputo
fion. Furthermore, the effecfiveness of the Waigel penolfies is untested and in the cose
deterrence foils, the actual imposition of the fines will oggrovote α fiscal crisis rather
than alleviate it. For these reasons the stability pact is at most α partial substitute for the
convergence criteria.
Co-ordinating convergence
Turning to the second principal role of the Maastricht criteria, in order to produce the
desired smooth transition to EMU α co-ordination of individual convergence efforts is
required. In fact each country's incentives depend on what other countries ore doing.
Apart from the usual policy spillovers from hscol and monetary policies, the various coun
tries' strategies ore interdependent via the probability (or fiming) of EMU which will be
0 function of joint effori. Consider two identical Agent countries who maximize utility as
before in Equation 3 above.
i ,3 Puo,;b
Volumel Noi 1996 89
(5) U(A)=p(E,E,)-T-^E'
Note that in equation 5 the entry probability is also α function of foreign convergence
effort. In principle the externality could be positive or negative. The most stroightforword
interpretation derives from the simple fact that EMU only happens if at least two coun
tries (often more precisely identihed as France and Germany) moke the Maastricht
appointment. In generol the chances of being admitted to EMU depend on various eco
nomic and political considerofions involving partner countries. For example, α
Maastricht induced recession next door lowers one's own probability of meeting the cri
terio. If an important trading partner looks like jumping the hurdle, one's own efforts will
intensify in order not fo be (eft behind. If other large countries stay out, the political sfigmo
of exclusion is reduced, and vice verso (witness the acceleration of Spanish and Italian
efforts in mid 1996).
If the start of EMU is conditional on ο minimal size requirement (even if that is not in the
treaty), again individual convergence which raises the probability of meeting the crite
ria has public good features. This is because the overall probability that EMU will go
ahead os planned is increased and therefore the expected payoff for oil partner coun
tries. This in turn increases the incentive to converge for everybody.
Consider Figure 4 for α simplified illustrofion of two Nosh equilibria, where jointly high
effort is assumed to leod to EMU for sure (p=1). Other parameter values ore as before.
Figure 4. The convergence game
Abroad
Home
In Figure 4, a country's best response, if no-one else converges, is to do nothing either
(bottom-right). Conversely, the greater foreign effort, the greater is the home incenfive to
converge (top-left). If countries start out in α low convergence equilibrium they will not
moke it to the EMU equilibrium without α co-ordination and commitment device to ini
tiate and support the tronsifion. This suggests thot α market-led or voluntoristic approach,
which advocates proceeding to EMU "when the fime is ripe" and convergence suh
ficient, is doomed to failure. The key commitment device that the Maastricht Treaty has
furnished to overcome this "horizontal" co-ordinofion problem (os well os the "vertical"
one between Principal and Agent) is to set both convergence requirements and α firm
deadline. Fixing α deadline and not imposing α minimum size on EMU should render at
High Convergence: £ = 1
£=0
High Convergence: (Ef=])
3,3
0,-1
(E, =o; - 1 , 0
0 , 0
E l B P o o c r s
90 Volumel N o i 1996
least some foreign entry probability close to one and thus provide incenfives for other coun
tries to catch up. Other measures to overcome co-ordinafion failure include the following.
Pivotal countries could set the standard and go ahead unilaterally (move order, mulfi-speed
EMU), mechanisms of co-ordinafion, communicotion and authority con be installed in
order to invoke the good equilibrium (e.g. the convergence reports by the Commission,
EMI, the EU summit declarofions etc.) and, finally, external commitment should help (e.g.
the German court ruling (6), the Bundesbank setting stability stondords).
The model of co-ordi
nation failure explains
w h y countries left i t
very late before they
init iated meaningful
convergence pro
grammes. In the pres
ence of uncertainty
about EMU'S fate i t
was rat ional to sit a n d
w a i t , as long as other
countries d i d the
same.
However, the co-ordination problem resurfaces if the treaty itself locks credibility. First, it is
politically unreolisfic and economically meaningless to conceive of α mini-EMU, especially
one that were to exclude either France or Germany. Second (and therefore), given conver
gence condifions os of 1 996 either the deadline or the strict interpretofion of the criteria
hove to give. As long as ο delay, ο failure of EMU or α relaxation of the entry conditions
ore perceived os possibilities, the model of co-ordination failure applies os it stands. In
porficulor, it can explain why countries left it unfil very lote, unfil many years after the
signing of the Maastricht Treaty, before they initiated meaningful convergence pro-
grommes. In the presence of uncertainty about EMU's fate if was rational to sit and wait,
especially as long as other countries did the some.
Given the lack of full credibility of the numerical convergence criteria, moreover, the
entry conditions as on incentive device in reality operate much more like relative than
absolute performance confrocts. For example, France is unlikely to try push its deficit
below the 3% limit in 1997 if it predicts that Germony will not meet the forget either.
Moreover, the inflation and the interest rate criteria ore explicitly relative condifions.
How strict they turn out to be depends on the behaviour of the three best-performing
countries.
There is ο further interpretation of the convergence gome of Figure 4 if the home coun
try is playing ogoinst the financial market rather than other countries. Then E would
denote the markets' levels of conhdence in the home country. If fhe market has optimis
tic expecfofions, infiation expectations and interest rotes foil and therefore also the hscol
burden. This facilitates convergence, the opfimistic expectations thus become self-ful
filling and the "good" Nash equilibrium is realized. Conversely, under pessimisfic mar
ket expectations (here f=0), α vicious cycle ensues and respecting the Maastricht cri
teria becomes difficult or impossible (7).
6} The German supreme court (Bundesverfassungsgericht) ruling in 1992 a n d a resolution of the Bundestag
insisted on a strict interpretation of the criteria and argued that Germany was only committed to join (or stay
in) a monetory union that safeguards price stability.
7) Examples of such multiple expectational equilibria include Calvo's (1988) model on debt default, Obstfeld
(1994) on speculative attacks, Eichengreen a n d Wyptosz (1993) on the EMS crisis.
E Ή l 'aocs
Volumel N o i 1 9 9 6 91
Building reputation for EMU
In order to explore the "signalling function" of the criteria recall the original formulation
of the "veriicol" game between Principal and Agent os given in equations I and 2.
Imagine that the Principal does not know the preferences of the Agent, in particular he
may be unsure about the β in equation 2. If β is high it is very costly for the Agent to
produce stability. Thus his joining EMU could undermine performance, e.g. lead to
higher inflation, in stage three. The Principal's payoff in stage three, therefore, con be
rewritten os follows (assuming 0 = 1 ) .
(6) V (P) ^p(Z + ωΕ(β))
The degree of convergence and stability in stage three depends negofively on the size
of β and therefore the Principol has on interest to prevent countries with α high β from
joining. Under complete informafion he would simply set convergence criteria strict
enough such that those countries would hnd it too costly to sofisfy them. At the some time
the entry barrier must be low enough as not to deter countries with α low β . However,
it may not be possible to separate the two groups if the low stability countries hove α
stronger incenfive to join EMU, i.e. α higher T,, in equation 2, and if the Principal is not
allowed to discriminate ogoinst porficulor countries, even if he knows that their entry
would jeopardize EMU performance.
If the Principal is not certain about candidate countries' preferences and stability orien
tofion, then high inflation countries may want to imitate the behaviour of low inflation
countries in order to gain admission to EMU. Conversely, low inflotion countries hove on
incentive to signal their type, i.e. choose actions that α high infiofion country would not
wont to follow. As shown in Winkler (1995) making entry to EMU conditional on satis
fying convergence criteria con be useful to seporote out high-infiofion countries and pre
vent them from joining. In this case uncertainty about preferences is resolved ahead of
EMU and stage three starts with α good reputofion and low inflation expectafions. In the
event that the two types of countries cannot be separated and both enter EMU, there will
be greater uncertainty about what policy the ECB will follow and its reputation will be
lower, inflation expectations and interest rotes higher. Even in this cose the criteria are
benehciol, however, since they induce lower inflation in stage two. This is all the more
important because in the run-up to EMU reputatìonol incenfives diminish for nafionol policy
makers. They will face α hnite horizon ("endgame problem") after which they will no longer
carry responsibility for monetary policy, and therefore con no longer be "punished" for bad
behaviour
E:B !\IDO:S
9 2 Volumel N o i 1 9 9 6
The role for stage two of EMU and of the convergence criteria in such α setting is to induce
and help candidate countries to convince the Principal and the morkets of their stability
orientatìon. Thus the convergence criteria con have on important funcfion with respect to
information revelation even if the behaviour they induce seems utterly poinHess and
destructive, as has been argued by many critics. In ο nutshell, candidate countries for
EMU ploy the role of the groom who has to woo α scepficol bride (Principal and finan
cial markets) before the EMU marriage. Conversely, the bride devises α set of tough exams
and obstacles in order to convince herself of the groom's serious and honest intentions.
France's dogged adherence to the "franc fori" policy ogoinst most economic advice is
α prime example of α signalling and reputation building strategy. Similarly the latter day
ERM managed to hold together and discipline countries with quite disparate stability tro-
difions, of least until the hnol reward of EMU (and hence the incenfive to converge) was
suddenly thrown into doubt with the Danish and French referenda in 1992.
The c o n v e r g e n c e
c r i t e r i a c a n h a v e a n
i m p o r t a n t f u n c t i o n
w i t h r e s p e c t t o i n f o r
m a t i o n r e v e l a t i o n
e v e n i f t h e b e h a v i o u r
t h e y i n d u c e s e e m s
u t t e r l y p o i n t l e s s a n d
d e s t r u c t i v e .
4 . Conclus ions
A major, recurrent crificism of the Maastricht Treaty regards the long "risky" transition
period of stage two. Indeed the obvious way to maximize the probability of EMU is to
keep the transition phase as short as possible, i.e. proceed to EMU quickly. As the paper
points out, however, the tronsifion is there for α reason: (prior) convergence is α condi
tion for the Principal's participation, treaty commitment is necessary to induce and co
ordinate prior convergence, and behaviour in the tronsifion period may reveal important
informafion. Moreover, if shocks can so easily knock EMU off course in the transition, then
perhaps either the economic condifions and the net benefits from EMU or the political
commitment ore insufficient and therefore it would indeed be unwise to proceed with
EMU. Also in these two dimensions the transition is on important testing ground and,
while risky, certainly not superfluous.
The paper proposed to interpret the Maastricht Treaty and its convergence criterio os α
contract device that seeks to organize the difficult tronsifion to EMU. It does so by de
termining the authority, timing and procedures for decision making ond by providing
rules and sanctions for behaviour. The treaty is certainly not ο perfect contract and many
of its details remain debatable. However, given its difficult and multifold tasks it moy well
be "perfectly imperfect".
The paper argues thot some entry conditions were required to reconcile the interest of
the contracting parties. It does not address the issue of whether the particular criteria
chosen in Maastricht is necessarily the best possible ones. It is, however, fair to point out
that any such criteria would have hod to be simple, transparent and non-discriminatory
in order to be included in on internofionol treoty.
Volumel Noi 1996 93
I f c o u n t r i e s a l r e a d y i n
t h e c l u b c o - o p t n e w
m e m b e r s , u n i o n s ize
c o u l d r e m a i n i n e f f i c
i e n t l y s m a l l a n d a
d i v i s i o n o f a t w o - t i e r
E u r o p e c o u l d b e c o m e
p e r m a n e n t .
As for the main policy questions, the Waigel proposal for α stability poet has already
been discussed. It should strengthen incentives in stage three and may allow ο limited
trade-off ogoinst the entry criteria. We con also use the framework to analyse the pros
pect of mulfi-speed EMU, i.e. the opfion of staggered entry. One crucial question
concerns who will decide on entry. If countries already in the club co-opt new members,
as assumed in Alesino ond Grilli (1993), union size could remain inefficiently small and
α division of α two-tier Europe could become permanent. However, the treaty does not
give the power of co-optation to the insiders of the EMU club, but allows countries "in
derogation" to join once they are eligible. The important point therefore is to maintoin
sufhcient convergence incentives for the "pre-ins" who do not moke the hrst round. In
principle, their situation and incentives ore unchanged from stage two, except that the
penalty of continued exclusion may be even larger and on appropriately designed ERM
II should help in maintaining the momentum of convergence. Therefore the risks of α per
manent division of Europe con be limited.
On the other hand, the merits of α multi-speed approach ore, hrst, to allow more flexible
convergence horizons geared towards individuol countries' needs. Second, it helps to
minimize uncertainty about EMU performance, in particular os regards candidate coun
tries' willingness and ability to support price stability. A small ECB is more likely to inhe
rit Bundesbank reputation intact and con then build up its own credibility before odmit-
fing more "risky" candidates into the club. The central features of the Maastricht Treaty
explored in this paper, condifionolity paired with α deadline, mode α multi-speed EMU
if not inevitable, then the most likely outcome.
9 4
E'E Ρ α ο ί " .
V o l j m e l N o i 1 9 9 6
References:
Alesino, Α. and V. Grilli (1993). "On the Feasibility of α One- or Multi-speed European
Monetary Union". Economics and Politics ,5, pp. 145-159.
Backus, D. and J. Driffill (1985). "Rational Expectations and Policy Credibility Following
α Change in Regime". Review of Economic Studies, 52, pp. 21 1-222.
Bini-Smoghi, L., Τ Podoo-Schioppa and E Papodio (1994). "The Transition to EMU in
fhe Maastricht Treaty". Princeton Essays in International Finance 194.
Buiter, W.H., G. Corseffi and N. Roubini (1993). "Excessive Deficits: Sense and Non-
sense in the Treaty of Maastricht". Economic Policy, 16, April, pp. 57-100.
Calvo, G. (1988). "Servicing the Public Debt: The Role of Expectations".
American Economic Review, 78, pp. 647-661.
De Grauwe, Ρ (1993). "The Polificol Economy of Monetary Union in Europe".
The World Economy, 16, pp. 653-661.
De Grauwe, Ρ (1994). "Towards European Monetary Union without fhe EMS".
Economic Policy, 18, pp. 149-174.
De Grauwe, P. (1995). "The Economics of Convergence Towards Monetary Union in
Europe". CEPR Discussion Paper 1213.
Eichengreen, B. and C. Wyplosz (1993). "The Unstable EMS".
Brookings Papers on Economic Activity, 1, pp. 51-143.
Hart, O. and B. Holmström (1987). "The Theory of Contracts". In: Bewley, TR (ed.)
Advances in Economic Theory: Fifth World Congress.
Cambridge, Cambridge University Press; pp. 71-155.
Kees, A. (1 992). "Maastricht - Vorentscheidung für eine Europäische Währung".
In: L. Grämlich, A. Weberand E Zehetner (eds.)
Auf dem Wege zur Europäischen Währungsunion.
Boden-Boden, Nomos; pp. 19-36.
Milgrom, Ρ and J. Roberts (1992). Economics, Organization and Management.
Hemel Hompsteod, Harvester Wheotsheof.
Myerson, R. and M. Sotterthwaite (1983). "Efhcient Mechanisms for Bilateral Trading".
Journal of Economic Theory, 28, pp. 265-81.
Obstfeld, M. (1994). "The Logic of Currency Crises". NBER Working Paper 4640.
; ,B »cj[:r:-'
Volumel Noi 1996 95
Rogoff, Κ. (1985). "The Opfimal Degree of Commitment to on Intermediate Target".
Quarteriy Journal of Economics, 100, pp. 1169-1189.
Walsh, C. (1995). "Optimal Contracts for Central Bonkers".
American Economic Review, 85, pp. 150-167.
Williamson, O. (1975). Markets and Hierarchies: Analysis and Antitrust Implications.
New York, Free Press.
Winkler, B. (1995). "Reputofion for EMU - An Economic Defence of the Maastricht
Criteria". European University Institute Working Paper ECO 9 5 / 1 8 .
Winkler, B. (1996a). "Towards α Strategic View on EMU - A Crificol Survey".
Journal of Public Policy, /6, pp. 1-28.
Winkler, B. (1 996b). "Of Sficks and Carrots - Incentives and the Maastricht Rood to EMU"
Forthcoming: European University Institute Working Paper (ECO].
t ' f i i^ClOC'S
96 Volumel Noi 1996
H o w to f i x conversion rotes at
the start of EMU
• ^ H -
^ m f Paul De Grauwe
^ Professor,
Katholieke
 ^ ' % . Universiteit, Leuven,
a n d Member o f
Parliament
1 . I n t roduc t i on
The problem of how the conversion rotes will be determined ot the stori of the monetory
union on January 1, 1999 (assuming that this will be the starting dote) is looming larger
OS we approach that dote. The problem is compounded by the fact that the decision
concerning membership in EMU should be token in early 1998, whereas the decision
about the conversion rotes con only be token on January 1, 1999. This creates α tran
sition period of uncertainty about the conversion rates ond may invite strong and des
tabilising speculation.
Broadly speaking one con solve this problem in two woys. A hrst method is to announce
the conversion rate ot the same moment os the decision on membership is token (eorly
1 998). The advantage of this approach is that if the market believes this announcement,
the transition period will be characterised by stability. The obvious drawback is that in
the absence of full credibility the announced conversion rotes may be tested by the specu
lators. As α result, large speculative attacks may destabilise the foreign exchange markets
during that period.
The L a m f a l u s s y r u l e
consists i n a n n o u n c i n g
Hiat the c o n v e r s i o n
r a t e w i l l b e a w e i g h t e d
a v e r a g e o f the m a r k e t
rates d u r i n g a g i v e n
p e r i o d b e f o r e J a n u a r y
1 9 9 9 .
The second method in selecting conversion rotes consists in not announcing the conver
sion rotes and letting the market decide. Thus, in this approach the market rote on
December 3 1 , 1998 would be the conversion rate on January 1, 1999. This method
would also imply on announcement (i.e. that the market will decide), and may thus suffer
from credibility (i.e. the market may not believe it). However, this announcement would
not invite speculative attacks, because centrol bonks do not hove to defend α porficulor
exchange rote. The problem with this method of selecfing conversion rotes is that the
market rotes may be volatile and may drift owoy from their fundamental value during
1998, so that the conversion rate occurs at "mis-oligned" levels.
The two methods just discussed ore in fact two extremes among α wide spectrum of inter
mediate methods. For example, instead of announcing α fixed conversion rate one may
announce α rule that will be followed to select the conversion rates. Or moving closer to
the market solution one may announce that the market rote will be the conversion rate
with the constraint that the market rote must remoin between α pre-onnounced fluctua
tion band (e.g. the 30% EMS-bond).
In this paper we analyse two different methods of deciding about the conversion rotes (2).
One, which we will call the Lamfalussy rule, consists in announcing that the conversion
rote will be α weighted overage of the market rotes during α given period before
Note: / om grateful to Filip Abraham, Yunus Aksoy, Claudia Costa, Guido De Bruyne, Hans Dewachter, Chris
Hurst, Jon Strand a n d participants of seminars at the Universities of Leuven a n d Oslo for their comments.
2} For a recent analysis of problems relating to the choice of the conversion rates, see Lehmentf 1996), a n d
SBC Warburg(1996). See also Spaventa(l 996) for a more general discussion.
E ;B Pooc'.s
Volumel N o i 1 9 9 6 97
January 1, 1999. We coll this the Lamfolussy rule because the president of the European
Monetary Insfitute has been the most prominent personality proposing the use of such α
rule.
The second rule that we will onolyse is on announcement that the conversion rote will
be 0 hxed rate (e.g. the central rote). We analyse this rule also to controst it with the
Lamfalussy rule. It should be stressed that both rules suffer from α credibility problem.
The market may hove doubts that the announced rule will effectively be used ot the time
of the conversion Ponuory 1, 1999). We will, therefore, proceed as follows. We will
first formally analyse both rules ossuming full credibility. In α second stage we study credi
bility problems associated with these rules.
2 . The L a m f a l u s s y r u l e
According to the Lomfolussy rule the conversion rate is announced to be α weighted ov
erage of the market exchange rotes during α period prior to the start of the union.
Specihcolly, Lamfalussy has proposed that the weighted overage of the market exchange
rotes during 1996, 1997 and 1998 should be used to calculate the conversion rotes
that will apply in 1999. In this section we formalise this rule. One issue that will arise
here is to what extent this rule should be applied retroactively.
Let us start from α well-known model of the exchange rote (see Mussa(l 976)), which we
write OS follows:
(1) S, = Z,-HbE,ASt,,
where S is the exchange rote in period t, Z, is the vector of exogenous variables that
affects the exchange rote. We will coll them the fundamental variables. These variables
also include political factors that the markets deem to be important to determine the
exchange rote in period t; Ε(Δ8^+ι is the expectation held in period t about period's t + l
exchange rote. It will be assumed that expectation are formed rationally.
We con also rewrite (1) as follows
(2) S, = l / ( l - H b ) Z , - H b / ( U b ) E t S „ i
or
(3) S, = γΖ, + δΕ,8,„
where δ = b/(l-Hb) <1 is the discount parameter with which the future exchange rote is
discounted.
EIB Poc:o-.s
9 8 Volumel N o i 1 9 9 6
We now apply equation (3) to model the exchange rote when it is known thot ot some
final dote the exchange rote will be irrevocably hxed using on overage of the market
rotes observed during α period preceding this dote. Let us assume that this period
extends over three years (3). We con then write
(4) 8 ι = γ Ζ , + 5 E , S 2
(5) S2 = γΖ2 -t- öEjSj
(6) S3 = γΖ3 -ι- 5E3S4
(7) S4 = α,S] + a2S2 -t- OC3S3
where α; is the weight given to the market exchange rate of period i, and Σ α ; = 1.
We now solve the model consisting of equations (4) - (7) for Sj using the rational expec
tations assumption. Thus, here we ask the quesfion what the exchange rote would be if
the Lamfalussy rule were announced and implemented in period 1. (In α second stage
we will onolyse the quesfion of how the exchange rote is affected if the rule is announced
in 0 given period (say period 2) and applied retroactively (from period 1 to 3)).
Repeated forward substitufion of the expectofionol terms allows us to solve for Sj (4).
This yields:
(8) S, = γ Ω , Ζ , + γ δ Ω 2 Ε , Ζ 2 + γδ^Ω3Ε,Ζ3
where
(9) Ω , = 1 - 1 « 3 δ . α 2 ^ ) ^ ^
1 - (α3δ + 0.2?) -l· α,δ^)
(10) Ω 2 = L·^ > 1 1 - ( t t jô + α2δ2^Η α , δ ^ )
1 ( Π ) Ω 3 =
1 - ( α , δ -ι- α,δ^-Η α,δ^)
From (8) we observe that S, depends on the fundamentals expected during the tronsi-
tion period (up to period 3). The interesting ospect of this solufion is that the fundomen-
3} Lamfalussy has proposed that the 1999 conversion rate should be an average over the three preceding years.
Presumably this would be formulated as a daily average over three years. Here we consider one per iod of time
to be a year. The analysis wi l l be extended to the case where the unit of time is a day (see section 4).
4) Solving the system (4}-(7) requires an inhnite amount of iterations despite the fact that there is a hnal date
after which the fundamentals play no role anymore. This is due to the fact that each time the hnal date (time 4)
is reached, the Lamfalussy rule "throws back" expectations to the past. This produces an "echo-effect" that conti
nues a d infinitum but dies down at each round.
E'BpGoi-s
Volumel N o i 1 9 9 6 9 9
tal variables ore multiplied by factors whose values exceed 1. In order to understand the
origin of this mulfiplier effect thot arises when the Lamfalussy rule is opplied, it is useful
to compare the solufion (8) with the solufion one would obtain if the exchange rote were
freely floating (5). We then obtain for Sj
(12) S, = γΖ, + γδΕ,Ζ2 + Ί ^ ^ Ε , Ζ , + γ Σ δ Έ ι Ζ ; , , i=3
f a c h t i m e a s h o c k
occurs rtiis c h a n g e s t h e
e x p e c t e d c o n v e r s i o n
r a t e , a n d " t h r o w s
b a c k " e x p e c t a t i o n s t o
t h e s t a r t i n g p o i n t . W e
c a n c o m p a r e th is p r o
cess t o a n e c h o - e f f e c t ,
w h e r e t h e c o n v e r s i o n
r a t e acts a s a r e f l e c t
i n g b a r r i e r .
In the hooting exchange rote regime oil the fundamental variables ore mulfiplied by the
coefficients γ(< 1) and the discount factor δ (< 1), so that γδ < 1. In addifion. In α floafing
exchange rote environment the expected fundamentals beyond period 3 also play α role
in determining the current exchange rote. Note, however, that as the future recedes, the
weights given to the future fundamentals progressively decline (γδ'> γδ'* ' ).
We con now contrast the exchange rote under the Lamfalussy rule with this free hoofing
exchange rote. First, the fundomentols during the transition period (periods 1 to 3) hove
an amplified effect on the current exchange rote compared to the free floafing regime.
The amplification effects ore measured by the terms Ω ; which oil exceed 1. Second,
these omplihcotion effects increase for fundamentals farther owoy in the future, i.e.
Ω3 > Ω2 > Ω ι . The result is that the effect of the fundamentals farther in the future
exceeds the effect of the fundamentals closer to the present. It can be shown that this
result is independent of the discount factor, i.e.
γδ2Ω3 >γδΩ2 > γ Ω |
This contrasts very much with the freely floating exchange rote solufion. Where do these
differences come from?
The omplihcotion effect can be explained as follows. (It con be useful to look ot the struc
ture of the model as presented in equations (4) to (7)). Suppose new information arrives
changing agents' expectations about Z3. Through the rational expectations process this
affects the current exchange rote S, ond oil the other periods' expected exchange rotes
(S2 and S3). This can be called the direct effect, which we also find in the free floating
model. The changes in all these (expected) exchonge rotes, however, affect the expected
conversion rote S4 which is ο weighted average of these exchange rotes. This in turn
changes the expected exchange rote in period 3, and therefore also all the preceding
ones, which changes the expected conversion rote again. This process goes on ad in-
hnitum. It converges, however, because of the discounfing foctor. Put differently, each
fime α shock occurs in on expected fundamental variable Z^ this chonges the expected
5} This is not the only regime with which the Lamfalussy rule should be compared. In a later section we com
pare the Lamfalussy regime with a regime where a f ixed conversion rate is announced beforehand.
1 0 0
EIB Pope's
Volumel N o i 1 9 9 6
conversion rote, and "throws back" expectations to the starting point. We can compare
this process to on echo-effect, where the conversion rote acts os α reflecfing barrier
ogoinst which expectations (the sound) ore bounced back.
It con now also be understood why this echo-effect is stronger when α shock occurs in
Z3 (close to the reflecting barrier) than when it occurs in Z2. In the latter case the shock
does not affect S3 in the first round, so that the effect on the conversion rote is less strong.
The omplihcotion (echo-) effect implicit in the Lamfalussy regime has the important impli
cation that unexpected changes in one of the fundamentals (news) during the hrst period
(the announcement period) will produce more variability of the current exchange rote
than compared to the freely floating regime. For example, if because of new informa
fion agents change their expectations of Z3, the impact on S^ becomes
(13) Δ5, = γδ2Ω3ΔΕ,Ζ3
this compares with the change in the exchange rote under α free float which is
(14) Δ8 | = γδ^ΔΕ, Z3
How important are these differences quantitofively? From an inspecfion of the amplih-
cation terms Ω| one con see that the addifional variability in the Lamfalussy regime con
be quite substonfiol. For example, if we set the weights a ; = 0.333, and the discount
factor δ = 0.9, we find
Ω,
Ω2
Ω3
= 2.3
= 3.7
= 5.3
We conclude that the omplihcation effects are likely to be substantial, adding α lot of
volafility in the exchange rote in period 1 when news occurs about one of the under
lying fundamentals.
The previous discussion makes clear that the use of the Lamfalussy rule to determine the
conversion rates is likely to moke the exchange rates quite volatile during the run-up to
monetory union. This volotility con be mitigoted, however, by starting the opplicafion of
the rule by surprise, i.e. by announcing that the rule will apply retroactively. If this con
be done successfully, one con hope that the volatility of the exchange rote will be miti
gated by the very fact that α port of the current exchange rote developments will be
influenced by the post, and known, exchange rote trends. We, therefore, now analyse
E;B Paot;-s
Volumel Noi 1996 101
the application of the Lamfalussy rule, assuming that this is done retroactively. In order
to onolyse this regime we return to equotions (4) to (7). We now coll period 2 the pres
ent (current) period. Therefore, period 1 is the post, and period 3 and 4 are the future
periods. Everything that has happened in period 1 is known with certainty. We now
solve for the exchange rote in period 2 (S2). The solution con be written os follows:
(15) 82 = γΨ2Ζ2-<-γδΨ3Ε2Ζ3-ι-α|δ2Τ35,
1 - (α3δ + α,δ^)
(17) % = 1 >1 1 - (α3δ + a2Ô2)
We find that the omplihcotion effects, which we now measure by the expressions Ψτ
and T3, ore smaller than the corresponding omplificofion effects when the rule is
applied without retroocfive force (equofion 8). Nevertheless, the omplihcotion effect
remains large. Using the some numerical example as before (a3 = 0.33 and δ = 0.9)
we hnd that
Ψ2= 1.6 and ^ 3 = 2.3.
Thus, even in the retro-active application of the rule the current and future fundamentals
ore likely to hove α signihcont effect on the current exchange rote. In secfion 4 we return
to this issue when we simulate the rule using days as our unit of time.
3 . A n n o u n c i n g α f i x e d c o n v e r s i o n r a t e
An alternative to the Lamfalussy rule consists in announcing α fixed exchange rote (the central
rote, or any other hxed number) os the future conversion rote. If this announcement is per
fectly credible (we come bock to the issue of credibility later) the solution for the exchonge
rate in period 1 con easily be found from equafions (4) to (7) and from setting
(7') S4=S*
where S° is the announced fixed conversion rate.
The solution for S| can then be written as follows
(18) 5ι = γΖ,-^γδE,Z2-^γδ2ElZ3-^δ3S*
EIB Pope.-s
102 Volumel Noi 1996
As con be seen, the exchange rote solufion does not suffer from the amplification effect
found in the Lamfalussy rule. In addition, as we come closer to the conversion fime, the
variability of the exchange rote must necessarily decline. For example in period 2 we
hove
(19)
and
(20)
S2 = γΖ2 -I- γδΕ2 Z3 -l· δ^5''
S3 = γΖ3 -ι- δ5''
It is useful to point out here that the Lamfalussy rule where the weight on post observo-
fions is set equal to 1 and on the present and future observations equal to zero is equiv
alent to the hxed conversion rate rule. For example, use equofion (15) and set «1 = 1,
a2= «3= 0. This yields equation (19).
During the first half of
the per iod, the
Lamfalussy rule pro
duces significantiy
more dai ly volat i l i ty
compared to the free
f loat ing regime.
4 . A simulation exercise
The analysis of the previous sections can be complemented by simulating the two regimes
for choosing the conversion rates. We proceed as follows. Instead of defining three periods,
we now opply the analysis to doily observofions. We consider 400 days. Each day
"news" arrives in the market changing the current and future expected fundomentols. The
news is modelled as α "white noise" stochostic process. This process was scaled such that
the exchange rotes always remain within the ERM bond of free fluctuation of 30%. The
weights a; ore redefined os daily weights. Given that there ore 400 days, a^ = 1/400.
The year/y discount rote is assumed to be 10% (yearly discount factor = 0.9)
We hrst apply the exercise to announcing the Lomfolussy rule without retroocfivity. Thus,
the outhorifies announce that the conversion rate which will be used ot the start of EMU,
will be the overage of the market rotes storfing ot the moment of the announcement. We
simulated this rule (using equation (8) reinterpreted to hold for 400 days) and compared
these simulofions with α system of free floafing (equofion (13)). We show ο few repre
sentative examples of such simulafions in figure 1 and 2.
We observe the following. During the hrst half of the period, the Lamfalussy rule pro
duces signihcontly more doily volatility compared to the free floating regime. In approxi
mately the middle of the sample period the volatility under the Lamfalussy rule declines
relative to the volafility of the free floot regime. At the end of the period, the Lamfalussy
rule produces α smooth convergence towards the announced conversion rote (the
moving average). This has to do with the fact that os we approach the conversion period,
the relative importance of shocks in the fundamentals in influencing the exchange rote
declines, whereas the importance of the overage of the past exchange rotes increases.
• • B i O , ; ; - s
Volumel Noi 1996 103
Since there is no surprise about these past exchange rotes anymore, the exchange rote
is increasingly stabilised ond smoothly converges towards the announced conversion
rote at the time of the conversion.
We document the differences in volafility in the two regimes (free float and the
Lamfolussy rule) in the following table 1. This shows the mean of the standard deviation
of the exchange rote under the two regimes for different sub-periods in 100 different
simulafions. We observe that during the first 200 days of the opplicofion of fhe
Lamfalussy rule the volafility of fhe exchange rote is considerably higher than under α
regime of freely flexible exchange rote. This has fo do with fhe amplihcotion effect de
scribed in the previous secfion. In the second period, however, fhe use of the Lamfalussy
rule signihcontly reduces volatility.
Table 1. Mean of standard deviofions of simulated exchange rote (100 simulafions)
period 1-200
Floafing Lamfalussy
1.4 2.2
period 201-400
Floafing Lamfalussy
1.4 0.6
Figure 1. Lamfalussy rule and floafing rotes compared
103
101 -
85
floating rate
Lamfalussy
moving aver
.|.1jg,l„l||.«.l,l» |[|. |. a3=lHiBsg*"g!!äaJMiiiaBgiHaiHMgii^^ τ - Vf
Cvl r- o co CD CD CM tn 03
CD •>- ra to 00 o
T - · *
c o LO CM CM c\j CM co ra
03 n (D Oi o IM r t <a
CM
m m Days
104
ElB PoDG's
Volumel Noi 1996
Figure 2. Lamfalussy rule and floafing rotes compared
120
115 -
110
105
100
95
1 90
85
. . t . . « · - - - · - • ι .χι—
Λ' V A .-'-l-r^ . / ' . / " • • •
floating rate
Lamfalussy
moving aver
ÎiiaiÎÎiiÎî!igajjaat'»**-'-*ii*!aMÎ^^^ Oi ••ί
ο « CO t ^ Οί -r-
σ> CM m OO 1- Tt ra (D 00 ο ra m ^ 1- ·>- CM CM CvJ CM
Days
o ra co <35 CM o CM ·* <o c» ra ra n ra ra
A second observation one con moke from hgures 1 and 2 is that the application of the
Lamfalussy rule con produce drift in the exchange rote which exceeds the drift obtained
in the free floafing regime. Figure 2 illustrates this. This has to do with fhe foct that in the
early phase of the application of fhe rule fhe excess volafility is built into the averaging
formula, producing extra drift. Note that this result does not olwoys hold. It depends on
the exact sequence of the stochastic shocks driving fhe exchange rote. We come bock
to this problem of drift later.
From the preceding discussion we conclude that the opplicofion of the Lamfalussy rule
without refro-ocfivify has great disadvantages. The results also suggest, however, that fhe
rule has stabilising properties, but that these appear only during fhe second half of fhe
simulofion period. This suggests that the rule should be applied with retro-ocfive force.
(In fact, Lamfalussy has proposed fo use this rule retroactively). We, therefore, analyse
fhe stabilising properfies of the rule when it is applied retroactively. We implement fhe
rule OS follows. On day 201 the authorities announce that fhe conversion rote will be
the doily average of fhe 400 days prior to conversion fime. Thus, of the moment of fhe
announcement 200 observofions needed fo compute the future conversion rote hove
already passed and ore, therefore, in the information set of the economic agents. We
moke fhe assumpfion that economic agents did not expect this announcement, so that
the latter arrives as α complete surprise.
E IB PoDO's
Volumel Noi 1996 105
In order to evaluate the stabilising properfies of fhe retroocfive Lamfalussy rule we will
compare it with fhe other rule analysed earlier. In this second rule, the authorifies
announce on day 201 that the conversion rote will be fhe (hxed) central rote. In both
coses we assume that fhe economic agents hove full confidence that this rule will be
applied at conversion fime. Thus, the degree of credibility is 100% in both coses.
We show some representative samples of fhe simulated exchange rote in hgures 3-6.
The simulations hove been run in such α way that during the first 200 periods, when no
rule is applied and agents ore unaware that α rule might be applied later, the exchange
rote is driven by the some stochosfic shocks in both regimes.
The results hove the following interpretation. The stabilising properties of both rules ore
comporoble. However, α more formal inspecfion of the volotility of the exchange rote
(from day 201 on) reveals that the fixed-convergence-rate-rule produces less volafility.
This is shown in fable 2 where we present fhe mean of fhe standard deviations of fhe
exchange rote in 100 simulafions of the two rules. We also computed another measure
of volatility, i.e. the roof mean squared deviofion between fhe simulated exchange rote
and the announced conversion rote.
Table 2. Measures of volafility of the simulated exchange rate
Lamfalussy rule Fixed conversion rote rule
standard roof mean standard root mean
deviation squared error deviation squared error
0.81 1.35 0.57 0.84
We hnd that fhe Lamfalussy rule leads to volafility which is approximately 5 0 % higher
than the hxed conversion rote rule.
A second observofion one con moke from the results of hgures 3 to 6 is that the
Lamfalussy rule, even when applied retroactively, con still produce α lot of drift. This is
mode clear in figure 5. This feature is likely to lead to the need for applying wide bonds
of fiuctuofions. We pursued this quesfion further by comparing α larger number of simu
lations of both rules. This is done in hgures 7 and 8, where we present 20 simulations
of both rules. We observe that fhe use of the Lamfalussy rule leads to α situafion in which
the conversion rate used of conversion fime lies in α relatively wide bond. This suggests
that the Lamfalussy rule will hove fo be occomponied by the use of wide bonds of fluc
tuation. This is not the cose with the hxed conversion rate. In this cose fhe exchange rote
always converges smoothly towards the unique hxed conversion rote. Thus, the hxed-
106 Volumel N o i 1996
conversion-rofe-rule could be implemented together wifh α gradual narrowing of the
bond of fluctuations.
The m a r k e t e x c h a n g e
r a t e is o n a d o w n w a r d
t r e n d , w h e r e a s t h e
a v e r a g e o f p a s t e x
c h a n g e r a t e s is s t i l l
m o v i n g u p . The a v e r
a g e o f t h e p a s t e x
c h a n g e r a t e s s u d d e n l y
b e c o m e s a d e t e r m i n a n t
o f t h e m a r k e t e x
c h a n g e r a t e a n d forces
a l a r g e u p w a r d s
a d j u s t m e n t .
A third observation to be mode from fhe results is the following. Both regimes lead fo
jumps ot the moment of the announcement of the conversion rule. This is not really sur
prising since fhe onnouncement is unexpected, and like any "news", leads to α jump in
the current exchange rote. There ore differences in fhe noture of fhe jump in fhe two
regimes, however. In fhe cose of the fixed conversion rule, fhe jump is always towards
fhe centrol rote. This result is due fo the fact that when fhe exchange rote has drifted
owoy from its central rote (normalised of 100) prior fo the announcement, the announce
ment (assumed fo be fully credible) forces the exchange rote on α path close enough fo
the central rote so that if con reach the central rote on conversion fime in α smooth way
(i.e. without 0 jump). This result is similar to the "smooth posfing" condifion obtained in
target zone models (Krugman(1991 ), Berfolo and Svensson(1993)).
In the case of fhe Lamfalussy rule, jumps occur in both directions, depending on where
the market rote is situated relafive to fhe moving average of the post exchonge rates.
We show on example of such ο jump offer the announcement of the Lamfalussy rule in
figure 9. At fhe moment of fhe announcement, fhe market exchange rote is on α down
ward trend, whereas the average of the post exchange rotes is sfili moving up. The over
age of the post exchange rates suddenly becomes α determinant of the market exchange
rote (it was not before), and therefore forces the market rote to make α lorge upwards
adjustment so os fo move on α path that will allow for α convergence towards the
moving overage of fhe exchange rotes. In figure 10 we show the simulation of the fixed-
conversion-rofe-rule assuming the some underlying disturbances in the fundamentals. The
contrast in the direcfion of fhe jump is striking.
It should be stressed that jumps ore not always os pronounced as in hgure 9. In order to
hnd out how prevalent these large jumps ore we mode 100 different simulafions of the
Lamfalussy rule and found that in 30 coses the jump (positive of negative) exceeded 2%.
We conclude that the use of the Lamfalussy rule con create significant and quite unexpected
turbulence in the morket of the moment of its announcement. This is certainly on unoffracfive
feature of the Lamfalussy rule. The fact that, at the moment of the announcement, fhe exchange
rote may be forced fo moke α relatively large jump because the post moving overage of the
exchange rote (on orfificiol construct) has been moving in α direction unrelated fo current
trends creotes orfificiol turbulence in the market. This problem of the Lamfalussy rule is also
mode clear in figure 8 which shows the results of 20 different simulofion. We observe that
after the announcement the spreod of the exchange rotes ocfuolly increoses.
t iB 'f'(i[)("s
Volumel Noi 1996 107
Figure 3. Lamfalussy rule (from day 201 on
1 0 6
r- m CD CM ·<!·
ra 0 5 CM
i n o ) r a N . T - L n c j ) r a r ~ - T - i o • « t c o c n T - T t c o c O f - r a c o o o T - T - T - c M C M C M C M r a r a r a r a
Days
Figure 4. Fixed conversion rate (from per iod 201 on)
106
104 -•
•<f
CM r>- ra CO 0 5 CM i n 0 0
O) T- ra CO CO ο ·.- τ - τ - τ - CM
OJ CM CO Oi en m
Days
108
EIB Popo's
Volumel Noi 1996
Figure 5. Lamfalussy rule (from day 201 on
102
i n c 3 > O T f ~ ~ T - u ) C 3 j r a i ^ - ' - m c » r a K » - t o c M ' ^ t i ^ c D C M ' ^ t D c n - . - x f c o c o T - r o c o o o
T - - . - - r - - r - o > i C M C M C M T O r a r a r a Days
Figure 6. Fixed conversion rote (from period 201 on)
I U Ä : -
ga
se ·
94
Inde
x CD
IV
)
9 0 -
Ski VW\y
-
-
-
• • .iijij-«Ìni..
h 1
#sSfl*A
Ιψ/ \
\h V V
lf.J|iìH i';B ffiE '·*
v^ / Kf'
jl.iiPlI
y \ . . y ' - ^ j ^
^
r—"
Ì '?aH#jagBSB» •=ί CM
1 ^ • ^
Ο I--
ra en
CD
— • r -
σ> ra
CM CD
i n 00 T ^
00 C3 CM
T -
ra cu
·.* i n CM
N h-CM
ο ο ra
ra CM ra
to • *
«
σ> CD m
CM ο·> ra
Days
;-!b ." ape-.s
Volumel Noi 1996 109
Figure 7. Fixed conversion rote rule (from day 201 on)
120
Figure 8. Lamfalussy rule (from day 201 on)
400
110 E13 Paces Volumel Noi 1996
9. Lamfalussy rule (from day 201 on
Figure 10. Fixed conversion rate (from period 201 on)
105 -
95 -i*Hs .iFiiiiteiianHiMBi • ^ h - o r a c D c n c M i n c o T - ' ^ r ^ o r a c o o j c M c M ' * > - c » ' - r a c D c o o r a i n r > - o c M ' d - c o a 5 , - 1 - T - T - c v i c j c g a i r a r a r a r a r a
Days
Volumel Noi 1996 111
5. Credible and non-credible rules
In the previous secfions we discussed two rules to fix the conversion rotes assuming that
fhe announcement of fhe rules is fully credible. One of the interesting aspects of the
results is thot the jumps in the exchange rote observed of the moment of the announce
ment of fhe rules occur because these announced rules ore fully credible, i.e. becouse
the market is 100% sure that fhe rules will be implemented. In this secfion we contrast
these results with the results obtained when the rules ore not credible. We concentrate
fhe analysis on fhe Lamfalussy rule.
Let us assume that the authorifies announce the Lamfalussy rule but that the market gives
this announcement α 0% credibility (i.e. if does not believe the rule will be implemented).
In that cose the rule has no influence on the market exchange rate. (We continue to assume
that fhe market believes that monetary union will start at the end of fhe simulation per
iod). The market exchange rote is then simply the free float solution, whereby speculo-
tors believe thot the market exchange rote will be used os the conversion rate. We
contrast the two regimes (fully credible and non-credible Lamfalussy rule) in hgures 1 1
and 1 2, where we present 20 simulofions of fhe two regimes assuming fhe some shocks
in the underlying fundamentals. We observe (not surprisingly) that no jumps occur in the
exchange rote of the moment of fhe announcement when the announcement has no
credibility. When the announcement is fully credible, jumps occur frequently. As a result,
of the moment of fhe announcement, turbulence in fhe market is in fact greater thon
under the free float (zero credibility) regime.
We also observe that of conversion time (day 400) the success of the fully credible
Lamfalussy rule to narrow fhe spread of fhe exchange rotes is limited. At conversion time
fhe difference between the lowest and fhe highest exchange rate amounts to approxi
mately 10% (while if is 15% in the free float regime). This does not appear to be α large
gain given that the start of the Lamfalussy regime is likely to be quite turbulent.
In reality credibility is rarely 100% or 0%. It is more likely that the credibility of the
announced rule lies somewhere between these two extremes. As α result, the movements
of fhe exchange rotes will lie between the two extremes shown in hgure 1 1 and 1 2. As
we move from 100% credibility towards less credibility the jumps of fhe moment of the
announcement, and therefore, turbulence, will tend to decline at the cost, however, of
more turbulence later.
EIB PoDO-s
112 Volumel Noi 1996
Figure 11. Lamfalussy rule
115
400
Figure 12. Free float
400
6 . E x p e c t e d a n n o u n c e m e n t
In the previous analysis we hove assumed that of the moment of fhe announcement of
how the conversion rates will be hxed, this announcement comes os α complete surprise.
Put differently, we assumed that prior to the announcement the market was unaware that
such an announcement would be made. This is quite on artificial assumption. The reason
is that when in early 1 998 α decision about EMU-membership will be mode the auth
orifies hove to moke some announcement: either an announcement will be mode about
EIB Papers
Volumel Noi 1996 113
how the conversion rotes will be hxed (e.g. α hxed conversion rote or α rule like the
Lamfalussy rule), or no such announcement will be mode. Therefore, before 1998, fhe
market will attach some probability that some announcement will be mode. Thus, the
announcement con never come os α complete surprise. This may then affect the dy
namics of the exchonge rote changes when fhe announcement is made.
In order to analyse this problem, we proceeded os follows. We assumed that prior to
day 2 0 1 , the market attaches equal probabilifies that fhe following onnouncements ore
mode:
• The outhorifies announce hxed conversion rotes (the central rotes)
• The authorifies announce the Lamfalussy rule wifh retroocfive force
• The authorities announce that fhe market exchange rote will be the conversion rote.
Clearly, other possibilities exist (e.g. no announcement of all). In order fo moke fhe ono-
lysis tractable, we limit the number of possible announcements.
The way we solve the model is os follows. For each expected announcement we hove α
solution for fhe exchange rote. When fhe market expects that the Lamfalussy rule will be
followed the solufion is given by equation (8); when it expects the fixed conversion rote
the solution is given by (1 8), and when it expects that the market rote will be fhe conver
sion rote the solufions is given by (12). Prior to day 2 0 1 , the market attaches the some
probability to these three announcements. Therefore, the solufion is α simple overage of
these three equafions (given that we ossume equal probabilities). From day 2 0 1 , the
solufion is given by either one of these three equafions depending on which announce
ment is mode. We show simulafions of the Lamfalussy rule and the hxed conversion rote
rule in hgures 13 ond 14. In both coses we hove ossumed, os before, that the under
lying stochosfic disturbances ore the some. We observe that the properties are very simi
lar to those reported in hgures 7 and 8. In both coses the exchange rote jumps ot the
moment of the announcement. These jumps, however, hove become somewhat smaller
under the Lamofolussy rule, although fhe difference is rather small. The some con be said
about the hxed conversion rate rule. Clearly, our assumpfion about the probobifity dis
tribution of fhe different announcements allows for α significant porfion of surprise. The
other results noted earlier ore also found here, i.e. the jumps under the Lamfalussy rule
ore frequenfiy owoy from the central rote. This contrasts with the regime of α hxed
conversion rote. Finally, we find α lot of drift in the exchange rote under the Lamfalussy
rule.
E'B '^ope's
114 Volumel Noi 1996
Figure 1 3. Fixed conversion rate rule (announcements expected)
us
uo
105
100 -
95
-i s 90
85
• j ^ ^ ^
^£^S ^ ^
-
1
r/y' ^ ^ ' ^
^ ^ ^ ^
^ ^ ^ ^ ^ ^ ^ Γ 3
g2p&^ ^^^Aîtsd
1 \
^ ^ ^ ^
^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ S * · ^ ^ ^ ^ ^ = " ^
1 1 1 1 1
50 100 150 200
Days 250 300 350 400
Figure 14. Lamfalussy rule (announcements expected)
120
400
'•'Β Oriri"S
Volumel Noi 1996 115
7 . C o n c l u s i o n : Issues o f c r e d i b i l i t y
The problem of choosing conversion rates arises from α frode-off between stability and
credibility. Announcing α hxed conversion rote only produces stability if it is fully credible.
Not announcing α conversion rote (or α conversion rule) solves the credibility problem ot
the cost of volotility. The proposals to tie the conversion rote to on overage of post exchange
rotes ore inspired by the need to hnd α compromise between stability and credibility.
We hove argued that the Lamfalussy rule produces more problems than it solves. It will
lead to quite unpredictable volafility of the moment of its announcement. In addition, it
leods to substantial drift in the exchange rote after the announcement, so that α large
amount of uncertainty will continue to exist about the precise value of the conversion
rotes ot the start of EMU. This will also necessitate the use of relatively large bonds of
fluctuation until the start of the monetary union.
The L a m f a l u s s y r u l e
p r o d u c e s m o r e p r o
b l e m s t h a n i t so lves. I t
w i l l l e a d t o q u i t e
u n p r e d i c t a b l e v o l a t i l i t y
a t t h e m o m e n t o f i ts
a n n o u n c e m e n t a n d i t
l e a d s t o s u b s t a n t i a l
d r i f t i n t h e e x c h a n g e
r a t e a f t e r t h e a n
n o u n c e m e n t .
Against this view, it could be argued that the Lamfalussy rule is inherently more credible
than 0 hxed conversion rote rule because it permits drift in the exchange rote in response
to changes in fundamental variables. Experience, however, suggests that much of the
drift in the exchange rotes is not related to fundamentals but to chartist behavior and
technical analysis. As α result, on averaging rule may produce α lot of orfificiol drift. It
is then not obvious o-priori that the Lamfalussy rule is inherently more credible than α
hxed conversion rote rule. We conclude thot the "compromise" that the Lamfalussy rule
appears to be is no serious olternofive. We ore bock at square one: the choice between
α hxed conversion rate announced ot the some moment of the decision about member
ship, or no announcement at oil.
The analysis of this paper allows us to formulate some qualitofive judgements about this
choice.
The announcement of the fixed conversion rotes has potenfiolly great advantages. If it
con be mode credible, it allows for α smooth tronsifion from the market exchange rotes
into the irrevocably hxed exchange rotes. This is due to the fact that, if credible, the
announced hxed convergence rotes will force the market rotes into on increasingly norrow
bond of fluctuation, which becomes "infinitely" small at conversion time. (We have
called this the "smooth-posting" condifion of α fully credible announced conversion rote).
Thus, the potenfial odvontoges of such α regime ore large. The quesfion then is how such
on announcement con be made credible so that it con produce its benehts?
In general, the answer is that α commitment technology should be set in place ot the
moment of the announcement, so that the market hrmly believes in the outhorifies' com-
116
EIB Pace's
Volumel N o i 1996
mitment to the set of hxed conversion rotes. As α minimum, it should be agreed that the
conversion rates cannot be changed except by unanimity. This will eliminate the temp-
tofion of individual members to engineer α "lost devaluofion". (It should be stressed that
this is olreody foreseen in the Maastricht Treaty which stipulates that the decisions about
the conversion rotes shall be mode by unanimity, (see art. 1091)).
A commitment tech
nology should he sef
in place a t the moment
of the announcement,
so that Hte market
Brmly believes in the
set of fixed conversion
rentes.
A second element in this commitment technology can be called "institutional front-
loading", i.e. of the moment of the announcement, some institufionol changes that would
normally only be implemented on January 1, 1999 should be put into practice. Thus,
monetary policies of the countries accepted in EMU should already be decided upon
jointly during 1998. Another port of this commitment technology consists in declaring
that each participating central bonk will supply its own money in unlimited amounts in
exchange for the currency under pressure. At the same fime the porficipating central
banks moke α commitment to target the money stock for the area os α whole. Such α
solemn declaration, if credible, con beat back any amount of speculation for the simple
reason that central bonks that support onother currency can create unlimited amounts of
their own currency to be sold in the market. Once this is known by speculotors, they will
not hnd it worthwhile to undertake α speculofive attack (6). At the same fime, specula
tors know that the torgefing of the money stock of the whole area mokes sure that the
intervention ocfivifies do not affect the global money stock since one currency is sold in
exchange of another one.
If the countries accepted into EMU fail to set up such (or α similar) commitment technol
ogy they will take α risk. In that cose it will be necessary to allow for α sufficient amount
of flexibility in the exchange rotes. The existing 2x15% fluctuafion margin should then
be maintained so os to absorb speculative shocks that may arise. The authorifies con
strengthen the stability of this fluctuation margin by declaring that the central rotes will
be the conversion rotes. Even if the speculators do not fully believe this (because of the
absence of α commitment technology), there would be sufhcient uncertainty about this,
so that large deviofions of the exchange rotes from the central rotes would create α high
risk of large losses for those speculators betting ogoinst α currency. This would help to
stabilise the exchange rote within the bond.
6} See Obstfe ldf l995} on this issue.
I IB PoDC's
Volumel N o i 1 9 9 6 117
References
Bortolo, G. and Svensson, L. (1993). "Stochosfic Devaluation Risk and the Empirical fit
of Target zone Models". Review of Economic Studies, 60, pp. 689-71 2.
Krugmon, P. (1991). "Target Zones and Exchange Rote Dynamics".
Quarterly Journal of Economics, 106, pp. 669-82.
Lehment, H. (1996). "Wie sollen die Wechselkurse in der Europäischen Währungsunion
festgelegt werden?". Die Weltwirtschaft, Heft 1, pp. 61-69.
Musso, M. (1976). "The Exchange Rate, the Balance of Payments and Monetory and
Fiscal Policies under α Regime of Controlled Floafing".
Scandinavian Journal of Economics, 78 (2), pp. 229-48.
Obstfeld, M. (1995). "Models of Currency Crises with Self-fulhlling Features".
NBER Working paper. No. 5285.
Spavento, L. (1996). "Out in the Cold? Outsiders and Insiders in 1999:
Feasible and Unfeasible Opfions". Quarterly Review, Bonco Nazionale
del Lovoro, Special Issue, pp. 122-137.
Warburg, SBC (1996). "Pre-onnounced Rules for Euro Conversion Rates?".
Foreign Exchange.
EIB Popo's
118 Volumel Noi 1996
A proposal to stabilise the value of the ECU
Luis Gonzalez-
Pacheco,
Senior Economist, Chief Economist's Department
1 . I n t r o d u c t i o n
ECU-denominoted financial instruments hove achieved on unexpected success - ot least
until 1992. Although not supported by any Central Bonk or Government, and only launched
in 1 9 8 1 , ECU eurobond transocfions ranked among the top 5 currencies in the early
1990s. One perceived advantage during the 1980s was that the basket dehnifion of
the ECU provided α cost-effecfive diversification. A second ottrocfion of the ECU which
assumed its importance only by the end of the 1980s, was EMU and, with it, conver
sion of ECU into Euro.
Alfred Steinherr
Chief economist
Ί Ε Ε Ψ •Γ
From 1990, the spread which appeared between the ECU exchange rote and the basket
proved that the values of both were not strictly related. This was widely realised during
the EMS crisis of 1992 when the spread widened (1).
The impetus that the Maastricht Treaty could have provided to the ECU market was dim
inished by two factors. The first was that the Treaty dashed oil hopes of α "parallel cur
rency" approach to EMU, that is, to promote, or at least allow the ECU to gradually
replace nafionol currencies and to grow into the currency of the EMU. This was α poss
ible alternofive strategy that at least some market participants hod favoured, or expected
(recall the "hard" ECU proposal). The second drowbock hurting the ECU market were
uncertoinfies about when the EMU would start, if at all, and on which condifions ECU
contracts would be converted into Euro.
This paper starts with α review of ECU market developments. Secfion 3 interprets the
evolution and volafility of the ECU exchange rote in relation to the basket. Section 4
proposes α simple model of ECU exchange rote determination showing that the ECU
value is determined much like any other exchange rote. Since ο smooth path to EMU
would beneht from renewed confidence in ECU instruments, we argue that stability of
the ECU volue at por with the basket would be beneficial both politically and to the
ECU market. We propose, therefore, α "linking mechanism" in Section 5, and show the
advantage of such α scheme.
2 . B a c k g r o u n d
The ECU was created at the end of 1978, together wifh the European Monetary System
(EMS). It was defined os α "basket currency", that is one ECU was equivalent in value
to 0 basket of the currencies of the European Economic Community (EEC). This is the
dehnition of the "Official ECU". Over time, the composifion of the ECU has changed,
including some currencies from new members of the EEC and readjusting the shares of
the other currencies (2).
1) See ECU Banking Association (1991).
2) Of course, the amount of currencies in the basket have changed in line with exchange rate realignments.
EiB Popes
Volumel Noi 1996 119
Since September 1989, with the inclusion of the Spanish Peseta and the Portuguese
Escudo in the Basket, the composition of the ECU has not been modified, as the
Moastricht Treaty has definifively frozen the basket. Even, when Austria, Sweden and
Finland joined the EU on 1 Jonuory 1995, their currencies were not included in the com
posifion of the "Ofhciol ECU".
Based on the definition of the "Official ECU", α "private ECU" developed in financial
markets. Its remarkable success is illustrated in Figure 1 by the evolufion of new bond
issues in ECU (3).
Figure 1. ECU bond issues 1981 - 1996
35
30
25
20
-,15 υ ÏÏ10 ο
M
C 5
= m 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 '
Years
C) 1996 data are only for the first half.
Figure 1 shows that issues increased steadily until 1 9 9 1 , to reach α maximum of ECU
35 billion. Thereafter, it has declined gradually to 15 billion in 1995, (the amount of
ECU 4 billion in 1996 is only for the first half of the year).
This evolufion is conhrmed by looking ot the net flow of ECU bonds (difference between
new issues and reimbursements). This is presented in Figure 2. The amount of net flow
of bonds increased up to 1 9 9 1 . In 1992 the net flow was still significantly positive, but
in 1993 ond 1994 it was almost zero. In 1995 and 1996 net flows were negofive.
3) For details on the development of ECU financial markets, see Steinherr and Girard (1991).
EIB Papers
120 Volumel Noi 1996
Figure 2. Net flow of ECU bonds 1986 - 1996
30
(·) 1996 data correspond to 30/06.
Issues i nc reased
s t e a d i l y u n t i l 1 9 9 1 .
Thereafter t hey decl ined
g r a d u a l l y a n d in 1995
a n d 1996 ne t f l o w s
v^ere nega t i ve .
Figure 3 shows that during the lost 1 2 years, the banking sector has hod on ECU credi
tor position vis-à-vis the non-bonking sector. Assets reached α maximum at the end of
1992 with on amount of ECU 64 billion, before declining to ECU 51 billion in the hrst
quarter of 1996. Liabilities followed α similar evolufion during this period, with α maxi
mum of ECU 35 billion at the end of 1992 and α decline to ECU 25 billion in the hrst
quarter of 1996. The net position has stabilised in α ronge between ECU 26 and ECU
31 billion since 1991, offer α period of increase from 1984 to 1988.
Figure 3. Internofional posifion in ECU of Bonks vis-à-vis the non-banking sector
84 85 86 87 88 89 90 Years
91 92 93 94 95 96 ·
I Assets # Liabilities Net position
Source: Bonk tor International Settlements. International Banking and Financial Markets Developments, various issues.
Despite the fact that the private ECU was not created nor controlled by any monetary
authority, the market value of the private ECU was close to the value of the Official ECU
t .ü '"apcrs
Volumel N o i 1996 121
A g r o u p o f m a j o r
E u r o p e a n b a n k s
a c c e p t e d t h e c o m m i t
m e n t o f c h a n g i n g p r i
v a t e ECU f o r b a s k e t
ECU a t close t o p a r .
H o w e v e r , a f t e r 1 9 8 8 ,
t h e " f u n g i b i l i t y o f t h e
E C U " w a s e f f e c t i v e l y
s u s p e n d e d .
(that is, the basket value) up to 1988. A group of major European bonks (the ECU clear
ing banks) accepted the commitment of changing private ECU for basket ECU at close
to par for settlements of balances within the ECU payments system.
This link between the private ECU and the basket ensured α narrow spread between
both for many years. However, after 1988, the existing link between the private ECU
and basket disappeared when ECU clearing bonks decided for operational reasons
(bonks were obliged to suppori one day exchange rate risk on ECU payments and hod
transaction costs of consfituting the basket) to discontinue settlement in basket units. The
"fungibility of the ECU" was effectively suspended.
As α result, the exchange rote of the ECU in terms of the basket (i. e. the number of "bas
kets" to be poid for one ECU) has fiuctuoted, according to the supply and demand of
the ECU, just like any other currency.
3 . E v o l u t i o n o f ECU e x c h a n g e r a t e
The market value of the ECU with respect to other currencies is quoted every day. With
the exchange rotes of the eleven currencies included in the basket, it is easy to obtain
the value of the basket (Ofhciol ECU) in terms of the private ECU, summing up the
exchange rote of the currencies weighted by their corresponding amount.
The evolution of the exchange rote of the ECU with respect to the basket con be observed
in Figure 4. This depicts the percentage deviation of the ECU from the basket exchange
rote (so-colled the ECU Delta), since January 1990.
Figure 4. Delta of the ECU. Daily data 1990 - 1996
%
2
1
0
-1
-2
-3
-4
1990 1991 1992 1993
Days
1994 1995 1996
Source: European Commission. DGII - ECU Unit.
122 EIB Papers
Volumel Noi 1996
The v a l u e o f t h e ECU
i n t e r m s o f t h e b a s k e t
f o l l o w e d a d o w n v i f a r d
p a t h u n t i l r e a c h i n g
a m a x i m u m
s p r e a d o f - 3 % .
The value of the ECU has deviated substantially from parity with the basket. The behav
iour of the delta can be grouped as follows:
• May 1990 - March 1991. The ECU appreciated with respect to the basket, reaching
α maximum posifive spread of 1% in January, 1991.
• April 1991 - August 1992. The exchange rote of the ECU presented α very small devio
fion from the Basket. In December 1991 the Maastricht Treaty was signed and the
conhdence in the stability of the EMS was solid.
• End of August 1992 - October 1992. The exchange rote of the ECU fell below the bas
ket, with α maximum spread of 2% in mid-September. After the first Danish referendum
in April 1992, rejecting the Maastricht Treaty, ond the announcement of the French
referendum in June, strong doubts about the likelihood of EMU provoked α crisis of
confidence on the EMS. Sterling and lira exited from the system, ond the peseta was
devalued. The exchange rote of the ECU suffered from this instability, but by the end
of October the negative spread hod once more disappeared.
• November 1992 - July 1993. A quiet period on the markets, during which the recently
adjusted currencies in the EMS showed α stable evolufion inside the intervention limits
of the system. At the summit of Edinburgh in December 1992, many exceptions were
accorded to Denmark for the Third Phase of the EMU, allowing for the possibility of
rotihcotion of the Moostricht Treoty in α second referendum, in May 1993.
• August 1993 - October 1993. There was α new period of heovy depreciofion of the
ECU, provoked by the second currency crisis of the EMS. During the month of July,
strong speculative attacks against some currencies (French Franc, Belgian Franc,
Danish Krone, Peseta and Escudo) obliged Central Bonks to intervene massively. On
2 August, the limits of the currencies in the EMS were enlarged to 15% from the pre
vious 6% (Peseta ond Escudo) and 2.25% (others). This decision, forced by markets
events, undermined conhdence in the ECU. The delta reached α negative volue of
1.5%.
• November 1993 - April 1994. ECU markets calmed with only minor spreads between
the value of the ECU and the basket.
• May 1994 - January 1996. The value of the ECU in terms of the basket followed α
downward path until reaching α maximum spread of -3% ot the end of January and
the hrst days of February, 1 996.
• February 1996 - September 1996. Increased optimism in EMU led to α continuous
reduction of the spread from -3% to only -0.6% by the end of September.
i 13 Pofxvs
Volumel Noi 1996 123
This doto is summarised in Table 1 with the annual mean and variance of the delta for each
of the last 6 years. The steady negative trend, plus increasing volatility, is clearly visible.
Toble 1. Delta of the ECU (Annual overages of daily observofions)
Year
1990
1991
1992
1993
1994
1995
1996
Mean
0.19
0.08
-0.14
-0.16
-0.27
-1.14
-1.62
Variance
0.048
0.053
0.102
0.079
0.034
0.356
0.422
St. Dev.
0.219
0.230
0.320
0.281
0.184
0.597
0.650
This brief description of the evolufion of the ECU reveals that its value is only weakly linked
to the volue of the basket. As there is no monetary authority controlling the supply and
demand of ECU and stabilising its parity, fhe exchange rate of the ECU relies mainly on
the confidence that the markets accord to the stability of the EMS and the hnol conver
sion of ECU to Euro at par.
As it is not possible to discard the risk of new periods of instability in the European cur
rency market before 1999, it would be helpful to link the ECU exchange rate to the bas
ket exchange rote. To see how such α link con be achieved, we model the ECU exchange
rote ond use this to simulate the effect of reintroducing α "limited fungibility".
4 . M o d e l l i n g t h e ECU-Basket e x c h a n g e r o t e .
In modelling the exchange rate evolution of the ECU with respect to the basket, we must
consider the ECU to be like any other currency and formulate the hnanciol equilibrium
relotionship between exchange rotes ond interest rotes (4). Dynamics is obtained from
the exogenous evolution of the interest rotes of the basket.
Interest rote parity for the doily evolution of the ECU exchonge rotes is written os:
(1) E, ( q , , , ] - q , = R[ '>-r["
where:
c]i = logarithm of the exchange rote of the ECU with respect to the basket. When one
ECU equals one basket, the value of q, will be zero.
4} See Steinherr (1994)
EIB Popers
124 Volumel Noi 1996
R'''= Continuous doily zero-coupon interest rote of the basket ot time, t, and with α
maturity of one day.
//'' = Confinuous doily zero-coupon interest rate of the ECU ot fime, t, and with α matur
ity of one day.
Interest rote parity implies that in equilibrium the expected return due to the variation of
the exchange rote of the ECU must be equal to the differenfiol return of the interest rotes
of the basket and the ECU, i.e. if the ECU has on interest rote greater than the basket,
the exchange rate of the ECU must depreciate with respect to the basket. Otherwise,
there would be riskless arbitrage opportunities.
For α complete description of the evolution of the exchange rote, we need to describe
the evolution of ECU and basket interest rotes. Clearly the interest rote of the bosket is
determined by the (exogenous) interest rotes ofthe currencies included in the basket. The
easiest way of computing this rote is to compound the interest rates of the currencies
weighted by shares in the basket.
The efficient markets hypothesis, implies:
(2) R["=Rr. ,^e^ '
That is, the exogenous basket rote follows α random walk, where ε, is on error term.
With the interest rote of the basket exogenous, we still hove two variables in equation
(1), the interest rote and the exchange rote of the ECU. Both cannot be determined by
the equation, in the sense that if the interest rote of the ECU increases automatically, the
expected exchange rote would be reduced by the some amount. For each value of the
interest rote of the ECU, there would exist an expected exchange rote satisfying the
equofion. However, the interest rote of the ECU is actually hxed by the markets (depending
on the demand and supply of ECUs). In the ECU cleoring system, the overnight rote
(Intervenfion rote) is hxed by the Bonk of Internofionol Settlements (BIS) on α basket basis
for the preceding day.
(3) r>/>=R'r,
With these three equations, we hove α model for the evolution of the ECU exchange rote.
Equofion (1) con be reformulated in the following way:
(4) qr.i = q,+ R'i'-n"+cc<'i,
EIB Papers
Volumel Noi 1996 125
which means that the exchange rote of the ECU ot fime, t-i-1, con differ from its expected
value only by the random term α',+ι, which is unknown at time, t. The efhcient markets
hypothesis applied to the exchange market implies that the markets use oil the informa
fion available concerning the evolufion of the exchange rote, process it and include this
informafion in the expected exchonge rate. Thus, differences between actual exchange
rote and the expected rote must be due to unexpected effects, which can be assumed to
be rondom and uncorreloted. In addition, we also assume the variance of the random
variable to be constant.
Therefore, we hove the following properties for the random variables:
(5) EAa<;U) = 0; V , { a < ' > J = a t , Cov(a'i> , a < " ] = 0; \f t ^ s
for any value k > 0
For the basket interest rotes, we also assume the efhcient markets hypothesis, so that:
(6) EAe<'L) = 0; V , ( e ? i J = a | ; Cov(e<;> ,εΙ> ] = 0; W t ^ s
Finally we must also assume that both errors for different dates ore uncorreloted:
(7) Cov(e[" , a i ' > ] = 0 ; i { t * s
These hypotheses hove been tested using daily doto for the exchange and intervenfion
rotes. The zero-mean hypothesis on the doily voriofions of interest and exchange rates
can be stotisficolly accepted. The autocorrelations of the errors, ε ' Ι ' and a' l ' ore also
near zero and both series ore uncorreloted os assumed in the model. Less validity con
be accorded to the hypothesis that variances ore constant, mainly due to the high dis
turbances which occurred in 1992 and 1993. During the calmer period from 1994 to
1 996, these variances hove been stable.
From above we con derive:
(8) E, (q,^^] = q,->- R['> - //' for ony value j > 0
Equofion (8) generalises (1) to accommodate any log and results from on iterative oppli
cofion of equations (1), (2) and (3) from period, t-i-j, to, t (for more details see
Appendix).
EIB Popers
126 Volumel Noi 1996
Equofion (8) implies that the expected exchange rote for any future period is equol to
the actual exchange rote plus the actual interest rotes differenfiol. Thus, the variation in
the exchange rotes is expected to compensate the differential of doily interest rates.
However, if there were α cumulated deviofion of the exchange rote at time, t , there is
no mechanism to correct this deviofion, as the expected movement will be just to com
pensate future changes in interest rotes.
Analysis of the relationship between the actual exchange rote and future exchange
rotes, together with the intervention rote, (see Appendix) suggests:
(9) i,., = ^, + ii?^-r/'^+X^-. + Z < .
The expected exchange
rate for a n y future
p e r i o d is equal to the
current exchange rate
plus the interest rate
differential.
We observe that qi^j is equal to the actual exchange rote plus the differenfiol of interest
rotes plus two random components reflecting respectively the variation of the interest rotes of
the basket,Ve'/^,, ond the voriofion of the exchange rote from its expected volueVa'/ i , .
If we take expectafions at time, t , in formula (9), we obtain directly formula (8), as all
the random components hove zero-mean at fime, t .
We can also obtoin the variance of the process, assuming that e^and a, ore uncorreloted
and with constant variances, o j and σ „ , respectively:
(10) V,i<7,,jj= α-1]·σΙ + j . a l
Expression (10) shows that, although oil futures exchange rotes hove the some expected
value, the variance around this common value increases linearly with the log between,
t , and the future dote.
The model con be viewed graphically in Figure 5, where we hove projected the expected
future exchange rotes values and the conhdence limits that ore determined by the model
from two dotes (1990 and 1995).
EIB Papers
Volumel Noi 1996 127
Figure 5. Expected exchange rotes and 95% conhdence limits
1990 1991 1992 1993
Days
1994 1995 1996
The preceding results con be summarised os follows:
fn case of an init ial
deviat ion, w e cannot
expect that the exchange
rate w i l l revert to zero.
The deviation of the exchange rote from the basket equals (in on opposite direcfion)
the interest rote differential between the basket and the ECU. Thus, if the market starts
from 0 delta near zero, (as in the first point projected in Figure 5) and with α small
differential between the interest rates of the basket and the ECU, we can expect the
exchange rate of ECU to stay near zero. But, in cose of on initial deviofion from zero
we cannot expect that the exchange rote will reveri to zero. This can be easily appreci
ated visually in the second point projected in Figure 5.
• The conhdence limits on the expected exchange rote increases with the log. Thus, the
conhdence limit widens rapidly. In other words, anything becomes possible, α fact not
contribufing to reassure ECU investors.
• Finally, this model has hxed α relationship between the doily interest rotes of the bas
ket and the ECU through the latter's "Intervenfion rote". The Intervention interest rote
could be set with the goal of reducing the spread between the ECU ond basket inter
est rotes. But, as in the cose of exchange rotes, there is an increasing variance of the
random component, so that the stability effect could be very weak.
These conclusions ore conhrmed by the long periods of negofive delto seen during the
lost two years.
5. Limited fungibi l i ty model
Making sure that the value of the ECU is close to the value of the basket requires on
effective mechanism that brings about equality. One such mechanism would be legal
equivalence, implying thot any ECU debt could be settled either in ECU or in basket
units. Legal equivalence between ECU and basket does not exist, although the
Commission has elaborated α proposol to this effect.
128 EIB Popers Volumel Noi 1996
w h a t else could be
done to prevent the
remote, but never
theless real , danger
of a n ECU meltdown?
The proposal is for
" l imi ted fungibi l i ty".
If legol equivalence does not exist, α private acceptance could achieve the some result.
This is, precisely, what the ECU Banking Association achieved until 1987 by allowing
settlement in either unit. Whot is interesting is that equivalence in only α segment of the
market is oil that is needed, in this cose, in the ECU payment system. If, for example, oil
ECU deposits were to be repaid in either ECU or basket - whichever is higher in value -
this would achieve exocfly the same purpose. As with two interconnected water basins,
one only needs to control the level in one to ensure the some level in both.
Another possibility of ensuring parity between ECU and basket would be central bonk
support. One or severol European Central Bonks would have to intervene when α cer
tain spread is reached.
Short of polificol commitment through central bank support and short of legal equival
ence, what else could be done, assuming (os we do) thot it is in the public interest to
prevent the remote, but nevertheless real, danger of on ECU meltdown?
There ore two things the ECU Banking Associofion (EBA) could do. Why the EBA?
Because the EBA manages whot comes closest to α "monetary function" by operating α
payments system with intro-doy credit, and because it regroups in its clearing system the
48 most important bonks in the ECU market.
One possibility would be to set the overnight interest rate in line with the delta. This pro
posol has already been adopted by the EBA, but has not yet been implemented. But,
even if implemented, it would suffer from α major drawback: in times of pressure on the
ECU exchange rote, the overnight rote would hove to rise to α level of several hundred
percentage points to be effective - whereas the EBA would only accept variations of α
few basis points (5).
The second possibility is α return to fungibility. But as the drawbacks in terms of costs
and value-day ore well-known, α return to the pre-1987 system would be unottroctive.
Hence the proposal of "limited-fungibility", for example once per month or per quarter.
In the remainder of this section, we analyse whether this "limited fungibility" could sto-
bilise the ECU market. The proposal we ore considering is to establish fungibility ot cer
tain dotes known in advance (for example the lost clearing day of the month). At such
dotes, creditors con choose to be paid either in ECU or in the basket if the delta is nega
tive, while debtors con choose payment when the delta is positive. Obviously, ot these
"fungibility" dotes, the exchange rote of the ECU cannot be less than the value of the
bosket (except for α small difference due to transaction costs). For other business days.
5) For the daily interest rate differential to compensate a 10 bp exchange rate difference would require an
annualised interest rate of some 43%.
EIB Papers
Volumel Noi 1996 129
the divergence of the exchange rotes can be expected intuitively to remain quite small
due to the perspective of the next "fungibility dote". This solufion would retain to α large
degree the odvontoge of the "permonent fungibility", but would reduce its shortcomings
in terms of cost and risk, os the number of fungibility dotes is small.
The v a r i a n c e is u p p e r
b o u n d e d a n d t h e
u p p e r b o u n d d e p e n d s
o n t h e l a g b e t v ^ e e n
" f u n g i b i l i t y d a t e s " .
In order to analyse the effects of the "limited fungibility", we modify the model of sec
tion 3 to include the fungibility condition. For reasons which will become clear below,
we neglect transocfion costs. If we coll, T, the number of days between two dotes of "fun
gibility" and consider the stariing dote 0 as one of the these dotes, then the following
condifion must be added to the model:
(11) c3O='?T=gfe.T=0, k = 0, 1,
We use condition that q-r = 0 together with equofion (9) and obtain for the period 0< t < T :
(12)
Therefore,
(13)
cii = [q,^R'r/r^ Σεί'ι+Σ«-.)
i\ì j i ; . E,(a^-^ ,] = - (q , ^R \ ' ' . r ' ;
Equofion (1 2) means that the error variable ot the "fungibility dote", a-}., depends on α
term known at fime, t, and α sum of random effects to appear between, t-i-1, and, T-1.
At time, t, the expected value of ßj ' . , compensates the delta at time t from the lost fun
gibility date. The random term will compensate the new deviations to be produced from,
t- i-1, to T-1 .
While deviofions in the basic model hod α permanent effect on exchange rotes, now the
deviofions ore mointoined just until the next dote of fungibility. From this dote onwards,
the process has zero-mean deviofion os the stochosfic process has been driven by parity.
Evaluating the variance of the exchange rote around its mean, os we hove done with
the basic model, ond proceeding iteratively (for detoils see Appendix) one obtains
14) V,(q,,,]<T.((jj+aè]
Comparing this result with the corresponding one in the basic model, we see that in this
cose the variance is upper bounded and the upper bound depends on the log between
two "fungibility dotes". Thus equation (14) represents α suitable criteria to establish the
period between "fungibility dotes", depending on the acceptable level of the delta.
130 EIB Papers
Volumel Noi 1996
Finally, condition (8) creates α link between the term structures of interest rates of the ECU
and of the basket. We know that on dates, L· • I, for i.'= 0, 1, 2, ... the exchange rote of
the basket and the exchange rote of the ECU will be the some. At these maturities, the
interest rotes on both currencies must also be the some to prevent arbitrage opportunities.
We hove then
(15) R ' y = r ' y ' f o r / : = 1,2
In the case of α different initial dote, t, with 0 < t < Γ, we con write:
q j - q, = fV-t] • (R,' ' - ή ' ' ] = -q, and os q ι = 0 we hove
(16) q, = ( T - t ] . ( r , " ' - R r ' ]
At moment, t, the values of R'/ ' and of r'/ ' ore known, so that q, is determined by equa
tion (16). This condition occurs on the next fungibility dote, determining the actual
exchonge rote. For the following fungibility dotes, the interest rotes of the ECU ore deter
mined by the interest rote of the basket for the some maturity ond the current delta.
With the actual differential of interest rotes, equation (16) imposes α very narrow bond
for the doto. In 1996 the meon differential of the short-term annualised interest rotes has
been 15 bp with ο standard deviation of 10 bp. Thus, with α probability of 97.5%,
we con expect that the interest rote differential will be less than 35 bp, or 3 bp on α
monthly basis. Adding transaction costs of 30 bp, we can expect that the maximum
delta with monthly fungibility must be less than 35 bp.
To compare both models, with and without limited fungibility, α simulation of α series
generated by each model is represented in Figure 6.
300 360 420 480
Without fungibility With monthly fungibility
EIB Papers
Volumel N o i 1 9 9 6 131
As both processes stari ot zero, the expected long term value of the delta for both will
be also zero. However, the hrst one allows for long periods of large deltas, while the
second keeps the series close to zero and strongly reduces the variance.
The some reasoning can be applied to the announcement of EMU which con be con
sidered OS 0 process with only one dote of fungibility (6). Certainty of EMU and taking
account of the fact that, on this dote, the ECU will be exchonged for on Euro ot the bas
ket value, would be sufficient, according to this model, to constrain the delta to α maxi
mum. As shown by equation (16) the maximum possible delta depends on the date of
EMU and the actual interest rote differenfiol up to that dote.
The impact role of EMU is arguably the reason for the recent decline of the delta. Since
March 1996, EMS exchange markets hove been calm and there has been α growing
conhrmation of the EMU calendar. This fact, in the absence of short term disturbances,
hos acted as on anchor pushing the ECU to bosket value.
6 . C o n c l u s i o n s
Despite the simplicity of the analysis three robust results emerge:
• The condifion of periodic fungibility would force the exchange rote to return to pority ot
the fungibility dotes, thus eliminating the delta. After each such dote, the process starts
again from parity, so the effect of any previous disturbance on the exchange rote of the
ECU is erased. In between fungibility dotes, the delta con be different from zero, but its
value will be confined to the expected monthly variation of the delta (in the cose of monthly
fungibility). With the doto available, the monthly mean delta would be less than 0.01%,
which compares with the historical values of Table 1, ranging from 0.19% to -1.62%.
• With fungibility, the maximum variance of the exchange rote is bounded and depends
on the intrinsic doily variance of interest rates and exchange rates, and on the period
icity of the fungibility dotes. More frequent fungibility reduces the maximum variance
proportionolly.
• Interest rotes of the basket and the ECU will be equal ot the dates of fungibility. In be
tween, small divergences con occur, due to the effect of the Intervenfion Rate mech
anism. For the hrst fungibility dote, that implies α quick decline in the delta, which will
be dependent only on the interest rote differential. For longer moturifies, this condition
forces the interest rotes of the ECU and of the basket to converge.
We therefore conclude that the "limited fungibility" condition establishes α link between
the ECU and the basket which is strong enough to stabilise the exchange rote of the ECU
6} See Steinherr (1994) and Folkerts-Londau (1991).
EIB Popers
132 Volumel Not 1996
The condi t ion o f per iodic
f u n g i b i l i t y w o u l d force
t h e e x c h a n g e r a t e t o
r e t u r n t o p a r i t y a t the
f u n g i b i l i t y d a t e s , t h u s
e l i m i n a t i n g t h e d e l t a .
The cost s h o u l d b e
m i n i m a l .
and reduce the value of the delta. This will increase conhdence in the ECU and streng
then its role in hnanciol markets.
The cost of limited fungibility should be minimal. In principle, the mere announcement
should push the ECU rote to pority. If not, ο cost equal to the delta plus transaction costs
on basket construcfion would be incurred by net debtors. At present (October 1996) the
delta is in α range of -20 to -40 bp, so that the total cost could amount to 40-60 bp,
including transaction costs. Transaction costs could be soved if, instead of delivery of the
basket, cosh settlement was accepted. Of course, if cosh settlement were to be accepted,
then there would be no reason to sfick to limited fungibility, whose only motivation is to
save transocfion costs. Having established permanent fungibility in this way would, of
course, maintain parity ot all times.
Are there gainers and losers in this proposal? The biggest gainer is surely the ECU mar
ket and net ECU asset holders. Households are long in ECU as is the banking sector os
α whole (see Figure 3). Net debtors ore mainly governments and, to on insignificant
amount, non-financial enterprises.
EIB Popers Volumel Noi 1996 133
References:
ECU Banking Association. (1991). "The ECU -To be or not to be α basket - That is the
quesfion". EBA Newsletter. Special Issue 6, June.
Steinherr, A. and Girard, J. (1991). "ECU financial markets: Recent evolufion and pers
pectives", in Steinherr, A. and Weiserbs, D. (eds.) "Evolution of the
International and Regional Monetary Systems" London: Macmillan.
Steinherr, A. (1994). "ECU interest and exchange rotes: the key role of EMU",
in Steinherr, A. (ed.) "European Monetary Integration" London: Longman.
Folkerts-Landou, D. (1991). "The determination of market and theoretical ECU exchange
rotes". Eurostat. Doc. n° 75 E.
EIB Papers
134 Volumel Noi 1996
Appendix Derivation of the models
The basic m o d e l
The evolufion of the exchange rote of the ECU with respect to the bosket has been modelled
with the following equafions:
(Al) E, (qt^t - q, ] = R',"-r'," Interest Rate Parity
(A2) R't"= R ' , V ε ί " Random Walk for doily interest rotes of
the Basket
(A3) r ," = R% Intervenfion rote.
In these formulas, q t , is the logarithm of the rofio of the ECU to the basket, i.e., the log
arithm of the number of basket units needed for buying one ECU in period, t, and, R,',
and, r, , ore doily continuous interest rotes of the basket and the ECU respectively.
As we hove dehned in Section 2, the delta of the ECU, δ,, is the percentage deviation
of the exchange rote of the ECU with respect to the basket.
c / With this notation, we hove q^ = log(l- i-öJ« / iQO/ or ihot, the delta in percentage
terms con be roughly approximated by q,.
From (A2) and (A3), we hove that:
(A4) Κ ί ! ; - < = Κ ^ ; - < . , = ε ί ! ; , f o r / > l
As we hove discussed in the main text, (A l) is equivalent to
(A5) qr.i = qt^R''-r',"+a['!,
Iterofing this formula backwards from, t-HJ, to, t ^ h l , and using (A4) we obtoin:
a •- Û + R ' " r ' " -\-rr"' - η •t P^'^ A - ^ ' " - η •μ //'^ 4- /V'' -i- i/'^ -i-rv ' -Ht+i - qt+i-l + r^t+i-l 't+j-l +'^t+i - qt+i-l + £t+i-l + " ( + / - qt+i-2 + ^i+i-2 + '^t+i-l+ ^t+i-l +(Xt+i =· · ·
From which,
(AÓ) <?.., = '7, + Αί'^-/;'+χε',^, + Σ < ' 1 = 1 ί = I
Taking expectafions of (A6) and taking account that, ε[ ' ' , and, a'l'', are zero-mean ran
dom variables:
(A7) E,(q,J=q,^R<,'>-r',"
EIB Popers
Volumel Noi 1996 135
The variance of qi^j con also be computed from A6, os both errors ore uncorreloted,
and with constant variances:
i= 1
(A8) VJ^,,,j= ν,(^ε%]-,¥,(j;^aZ]= ί/-ΐ;.σ^/.σ^ Ho.+o'j î= I
In terms of the delta, we obtain:
(A9) f , ( S t J " ô . - t - - ' ' ' ' ' where, //'^, and, 4 " , are the over-360
night interest rotes (discrete) of the basket and the ECU ot time, t .
As con be seen, the delta ot any future time will be corrected only by the effect of the
differential of interest rotes. It is maintained ot this value without reverfing to zero.
The l i m i t e d f u n g i b i l i t y m o d e l
The fungibility condition imposes that ot certain dotes the delta is zero. Denofing by, T ,
the fime between two such consecutive fungibility dotes and assuming time zero as one
of these dotes, we hove:
(AIO) q ^ = q j = q ^ _ ^ = 0 , k = 0, 1, ... .
For any intermediate dote, t < T , and according to (A6), q j c a n be expressed os:
Therefore:
(All) <=-ii,+Ri"-ri"j-(l:Wii+2:«i'i,) .= 1 1=1
We con thus split olj,, in two components: -(q, -i- R't' - r ' l ' ] , which is determined at time, ,T-I-I T-t-t .
t , and - l ^ e i + j - t - ^ ' ^ ' + i l - h ich is random at time, t , but will be completely deferi i ι 1=1
mined ot time, T - 1 .
This formula indicates that clj., will compensate the actual deviation (determinisfic com
ponent), plus all the future deviofions to be produced from, t , to, T-1 (random component).
In this way the fijngibility condition modifies the assumed chorocteristìcs for the variable, cr/., :
(Al 2) E, fai'.g =-fe, + Ri' -ri' j
EIB Popers 136 Volumel Noi 1996
(A13) V,(a [>J= (T-t-l].(al^aÌ]< (Τ-ή.(σ1^σΙ,]
Formula (Al 3) shows that the variance of the exchange rote is bounded and that it
depends on the time to the next fungibility date. Then its maximum value is reoched
immediately after α fungibility dote, and this maximum depends on the time between fun
gibility dates.
If in formula (A6) we consider the period, j >T-t , we obtain:
<?,., = <?, ^R l'- /;^+Σ^''^Σ«-'=('?' +^!"- i '+'èi 'i.+i«'-) +Σ^!'.+Σ«: ί=ι 1=1 /=i 1=1 (=τ-ι i=r-(+i
The expression between brackets is, q j , which is zero. Therefore:
/ - I
(A 14) ί,.,=Σ^ '' + Σ« t + i
T-t i = T-!+l
This equotion means that the history of the process is erased ot each fungibility dote and
the process starts again from zero. This is α major difference with the basic model,
where shocks hove α permanent effect.
Finally, for the interest rotes, we hove:
(A15) qk.T- qr φ.T-t] .(R<'•^-"- r,"-'--"]= 0
For k =1 this formula gives on arbitrage condifion for the actual exchange rote and the
interest rotes ot maturity, T :
(A16) q,+(T-t].(R:'--"-r:'-->i=0
This relofion is not exact in the real world becouse of market imperfecfions, such as trans
action costs. It is probably the reason for the current decline of the delta following increased
conhdence in the EMU.
For ^ > 1 , we get α condifion for the ECU interest rote:
(Al 7) r/*- -'' = R/fc-T-'V_^^
' ' ' k.T-t
Thus, OS "k " increases, the difference between the interest rotes of ECU and the bosket
must be reduced.
EIB Popers
Volumel Noi 1996 137
Book revievNT
W i n d s o f C h a n g e . Economic T r a n s i t i o n in C e n t r a l a n d Eastern Europe
Daniel Gros and Alfred Steinherr, Longman, London and New York, 5 4 4 pages, 1995.
At the fime of the 1 989 revolution in Eastern Europe, which brought about the foil of
communism, it was α commonploce observofion that one would seek in voin any indi
cation in the works of economists os to how α centrally-planned economy could be trans
formed into α morket economy. Much has changed in this respect in the intervening
years. The book I wish to discuss comprises 5 ports, each ending with α detailed bibli
ography which primarily lists books and oriicles about the tronsifion from the communist
system to α market economy.
It would not in fact surprise me if the book written by Daniel Gros and Alfred Steinherr
came to be regarded os α standard work in this held. The authors summarise their work
very concisely as follows:
"After the description of socialist economic organisation and performance in Port I and
the mainly conceptual discussion of the reform problems in Part II, the remainder of the
book concentrates on the concrete situation in reforming countries. The reforms already
carried out ore evaluated and compared, os ore the problems sfili to be solved. Port III
focuses on Central Europe and leaves the former Soviet Union (FSU) to Port IV because
the size and nature of the problems in the FSU ore incommensurable with those in
Central Europe" (page 231).
In ο work such os this it is α good idea to provide α fresh general analysis of commu
nism OS ο social ideology and of the socialist economic system. This is what Pari I does.
One of the mistakes made was that the best of everything went to the military-industrial
complex; this was then perhaps able to compete with the West but the remainder of the
economy suffered from α lock of compefition, protecfion from foreign influences and
insufficient access to Western technology, as α result of which the opportunities for growth
were limited. In the period after WoHd War II the lock of any kind of price mechanism
for the ollocofion of capital hod a growing impact. After the collapse of the system it
was also discovered that production in the Soviet Union was lower thon hod previously
been assumed. The "obsession with growth" is also discussed. It is pointed out earlier in
the book that autarchic tendencies ore not inherent in socialist principles and that the
lock of α proper cost allocation system resulted in large-scale producfion of goods with
α negative added value.
Translation of the Dutch original published in Maandschrift Economie, 6, December 1996.
EIB Papers
Volumel N o i 1 9 9 6 1 3 9
Part II deals with "tronsifion and reconstrucfion", discussing everything that will hove to
be done in Eastern Europe to create α social market economy. First the scale and the
chronology of the reforms are discussed, followed by price liberalisation, foreign trade
ond foreign exchonge liberalisofion, macro-economic stobilisofion, privofisofion and
reforms in the financial sector.
Eoch chapter ends with α number of conclusions. The authors stress, for example, that
the credibility of α reform programme and sufficient political support ore so important
that they hove to take precedence over the speed and comprehensiveness of the "big
bong". Another conclusion is that there is no point in liberalising prices unless finonciol
discipline is also introduced in enterprises. If subsidies confinue to be provided, pro
duction will not be adjusted. A measure which must be token to complement price lib
eralisation is liberalisofion of international .trade, which strengthens compefition.
Domesfic capital markets also need to be developed in this context. Gros and Steinherr
advocate free exchange rotes for capital transactions and α hxed rate for the current
account OS an interim solufion to the problems of international capital transactions. In the
cose of the Visegrad countries this is no longer necessary, though it is for Russia.
The authors hove some interesfing comments on the "sofety net" for those secfions of the
populofion who ore hit too hard by the reform programme. They ore in favour of income
transfers but prefer α "give-away privofisotion programme that provides cifizens with real
assets to compensate for the reduction in income streams" (page 175). Those assets con
then serve as security for loons in bod fimes. It is on interesfing idea because the state
is naturally going to sell off houses and shares in state-owned industries os port of the
privofisofion programme anyway. The authors believe their suggesfion is honest because
they assume that everyone will receive on equal amount and that difhcult valuation prob
lems will be avoided.
As I believe the safety net to be of crificol importance, I should like to make α few observ
ations on this. As α solution to the privotisotion programme, I find the suggestion attract
ive but I would quesfion whether it constitutes α "safety net". I do not think that the idea
of using the newly acquired assets os security for α bonk loon is α particularly good one
because it would mean people who were already not enjoying economic success would
incur debts which would hove to be repaid. The only opfion then open to them would
be to sell. That would offer some temporary relief but the people concerned would be
eofing into the assets they hod only just acquired. Personally, I believe that free support,
i.e. income transfers from the state, is unavoidable in the cose of the poorest and I should
hove liked to have read more about this aspect.
EIB Popers
140 Volumel Noi 1996
In fact, oll the countries concerned hove chosen α mixed programme: some important
enterprises remain in stote ownership, others are being sold (through α coupon system,
for example), while others again ore being closed down or placed provisionally under
the control of α trust ("Treuhand").
For some citizens the reform programme has serious consequences ond, after 7 years,
I am amazed that in Eastern Europe there was not much social unrest.
The second port of the book ends with on examination of the hnanciol sector. The moin
recommendation is, of course, that the central bank should be independent and that this
should be laid down in the country's constitufion. The central bank should also supervise
the banking system. The idea that the banking industry, which should comprise univers
al bonks, should be divided into three categories is on interesfing one; the hrst catego
ry would meet the highest requirements as regards solvency and would pay the lowest
premiums for deposit insurance and the lowest rediscount rate. The reserves these bonks
would be obliged to hold would also be the lowest. The third category would not be
covered by deposit insurance or rediscounting and would not participate in the pay
ments system.
I wonder whether this system would not simply constitute α death sentence for banks in
the third category. Who would wont to keep α deposit account with α bonk described os
having the lowest possible standing even though it offered higher deposit rates? It would
be better simply to prohibit bonks which do not meet clear requirements, including those
regarding the reliability of the directors. 1 ogree with the authors, however, that the evolv
ing regulofion of bonks by the EU could act as α good example to Eastern Europe.
Port III of the book deals with the actual reforms in Central Europe, beginning with the
transformafion of the GDR into on integral part of the Federal Republic of Germany. The
authors ore frank in their views: currency union was α correct decision and the exchange
rote of 1:1 is not responsible for the high unemployment in Eastern Germany. I should
like to odd that it would hove been impossible to explain to the public α conversion rote
of, for example, 2 : 1 . It should also be borne in mind that α different conversion rote
would in any cose hove been overtoken by wage rises. Gros and Steinherr ore not
enthusiastic about the Treuhand agency: α better solution to the privofisofion issue would
hove been simply to hand over the housing stock, rural properfies and small businesses
fo the East German people. The authors point out that income transfers hove ossumed
enormous proportions but advocate in this context α "marginal employment subsidy"
involving negofioble coupons for the unemployed, to be sold in return for work or train
ing. On the basis of the high level of investment in the former GDR, Gros and Steinherr
EIB Papers Volumel Noi 1996 141
onticipote that in 10 to 15 years' time the region will have caught up with Western
Germany to such on extent thot it will be producing 60-70% of GNP per capita in the West.
The following chapter concerns the transformation of the former Czechoslovakia,
Hungary and Polond and is based on contributions from α number of IMF officials. The
conclusions rightly state that: "The interplay of structural and macroeconomic problems
across the different sectors of the economy makes the formulofion of policy particularly
onerous, especially since these economies ore pioneers in facing the challenges of sys
temic transformation to the market" (page 314). The conclusions also stress the problem
of privofisofion in the light of the many, somefimes contradictory, objecfives of govern
ment policy. Attention is also devoted to the importance of good bankruptcy regulofions
and the burden imposed on the state by unemployment benefit, the need for which arises
from the dismissal of superfluous employees in industry. The authors olso worn us not to
forget the need for α complete overhaul of the tax system. Pending the development of
capital markets, the hnoncing of government deficits and company investment will largely
be the responsibility of the bonks, which must also be subject to proper supervision.
My overall conclusion from this chapter is that the Visegrad countries hove ochieved α
lot since the revolufion of 1989. Inflation, for example, has fallen sharply and the shore
of GDP accounted for by the private sector in Czechoslovakia increased from almost
zero in 1989 to 60% in 1994. In the cose of Hungary the corresponding hgures ore
16% and 5 5 % and in the cose of Poland 28-30% and 5 5 % (page 282).
The final chapter of Part III concerns the former Yugoslavia. It is impossible for it to be
up-to-date, in view of the civil war that has ravaged the country since 1991/92. One
would be wrong, however, to ignore this port of the book. Consideroble emphasis is
given to worker seff-monogement. An enterprise run in this way could theoretically even
be better than α capitalist enterprise but in order for this to be so the workers would hove
to hove tronsferoble rights of ownership and this was not the cose in Yugoslavia. If rights
of ownership are transferable there is on incentive to reinvest rather than creaming off
the profits in the form of cosh. Another mistake was the negofive real interest rote, with
oil its consequences for the allocation of the factors of production. There was also α liber
alisofion process which come to α holt holf-woy through, and like others before them,
the authors stress how the six republics were growing owoy from one another. The fact
that the governments of the individual republics hod far too little eye for their mutual
interdependence and pursued autarchic policies is seen os α major mistake.
Port IV of the book looks ot the former Soviet Union and its disintegration, α process
which reached α climax in 1991 (chapter 1 3), ond the initial attempts at reform ond sta
bilisation in Russia (chapter 14).
EIB Papers
1 4 2 Volumel N o i 1 9 9 6
As regards the first problem: the authors concentrate on two, in their opinion, extreme
views, namely that the new states within the territory of the Soviet Union should immedi
ately hove introduced nafionol currencies, or olternotively that these states were so in
tegrated that they should have retained monetary and economic union even though they
hod become politically independent.
The authors offer α few percepfive analyses, rendered all the more remarkable by the
fact that they are based on extraordinarily weak material. The hnding of one of the
analyses is that there was for more trade between the Soviet republics than would hove
been expected if free internofionol trade hod prevailed among them. Trade between
them would therefore become marginal if international markets were opened up.
As regards the possibility of α monetary union between the former republics, the following
sentence from the book is interesfing. "The ruble zone started to crumble once the Central
Bonk of the Soviet Union (called Gosbonk) started to lose control over its head offices loca
ted in the 15 republics" (page 382). This indicates therefore, that no central monetary poli
cy was established, and this would hove been α precondifion for monetary union.
Pending full convertibility between the currencies (ossuming thot no monetary union was
achieved) α clearing mechanism should have been established, like the Europeon
Payments Union in Western Europe in the first few years after the war. The authors point
out thot an Interstate Bonk was planned but that the agreement concerning it was never
implemented. The same is true of α Central Bonk Council for the Commonwealth of
Independent States.
Personally, I am inclined to opt for the second of the two above-mentioned extreme
views. That is to soy that when the republics became politically independent, α mon
etary union should hove been established, but, os stated above, with α common mon
etary policy. The starting point for participation in internottonol economic relations
would then hove been stronger than it in fact was. Chapter 14 reads depressingly. After
Russio hod separated from the other republics in December 1991 attempts were mode
to carry out reforms and achieve stabilisation. A few quotes will suffice to give on
impression of the many failures and disappointments:
"None of these programmes (1990-1991) could be implemented, however, os long os
President Gorbachev did not really believe in α market economy" (page 408).
"Toble 14.1 clearly shows that Russia hod by for the highest price jump among the group of
four economies with rapid price liberalisofion" (Russia, Poland, Bulgaria and
Czechoslovakia, page 410).
EIB Popers
Volumel Noi 1996 143
The following is extremely important: "The entire legal structure for α market economy
did not exist in Russia. In Central Europe, large ports of the pre-war legal system hod
survived, ot least on the books, and provided these countties with on acceptable start
ing point. In Russia nothing similar existed" (pages 414-415).
The authors ore poriicularly crificol of the fact that energy prices were not brought more
in line with internofional market prices and that the government continued to subsidise
import prices. One of the conclusions of this chapter is that since most of the mistakes
mode hove now been corrected, cautious optimism is appropriate (page 441). Given
what is said in the course of the chapter, I myself find this conclusion extremely opfimisfic.
Port IV goes on to onolyse what the rest of the world and Western Europe in porficulor
could do to support the reform process in Eostern Europe. Chapter 15 contains on
appendix with detailed hgures on the support actually provided. Gros and Steinherr ore
not happy: the hnanciol aid which hos been provided was too little and took the wrong
form (expori credit, for exomple). They believe that α kind of Marshall Plan with control
over disbursement would hove been and sfili would be better, parficulorly for the coun
tries of the former Soviet Union. The problem is: "without reforms no aid, without aid no
reforms" (page 474). However, the authors also believe that the aid needed is less than
is often estimated and that incomes in Eastern and Central Europe should be converted
in terms of equivalence of purchasing power rather than ot market exchange rotes.
The hnol chapter, chapter 16, concerns trade agreements and the enlargement of the
European Union. The two issues are linked: the authors believe that the Visegrad coun
tries (and Slovenia) hove already adjusted their patterns of trade to such on extent that
they could become members of the Union within α few years. A European Free Trade
Zone is advocated for the other countries, including those in the former Soviet Union.
This would comprise 700 million consumers and open the vast EU market to exports from
Eastern Europe. A separate appendix deals with arable and livestock farming. It culmin
ates in the following sentence; "Opening up its ogriculturol markets is arguably the
greatest contribution Western Europe con moke to Eastern European development"
(page 521).
Fans of internofionol trade issues will enjoy "Box 16.5", for example, which examines
and compares the structure of trade in the European Union and Eastern Europe. "Box
16.6. The opfimum size of α club", provides another interesting analysis. As elsewhere
in the book, there is mention of the importance of the "gravity model", which analyses
the significance of the proximity of countries ond regions for trade between them. The
results constantly stress the major importance of the European Union and more particu
larly Western Europe os α whole as the economic heart of Europe.
EIB Papers
144 Volumel Noi 1996
As regards the admission of the Visegrod countries to the European Union I should like
to moke one observation of α purely political nature.
When their book went to press Gros and Steinherr could not hove known that the posifion
of the Hungarian minority in Slovakia would deteriorate considerably as α result of
legislative changes. It is essential that events in this regord are monitored closely.
Economic considerations aside, the European Union should contain only countries where
minorities also feel ot home. States where this is not yet the cose could follow the
example of Belgium and the way it conducts relations with its German-speaking minority
in the east of the country.
I have deliberately summarised this major work on α chapter by chapter basis because
it is so rich in onolyses and suggestions that α shorter summary would not have been
0 fair reflection of it. All the problems ore discussed on the basis of solid principles of
economic theory using extensive stofisficol moteriol from sources including the former
Soviet Union, albeit that the latter is less reliable than that from the West. Each of the
four Parts contains on extensive bibliogrophy, but the literature concerned has also been
thoroughly incorporated in the analyses. The authors' conclusions, which ore given at
the end of each chapter, mostly come over as extremely logical and common-sense, but
it would do the outhors insufficient justice if one were to restrict oneself to those con
clusions and not look at the analyses.
In brief: this is α foscinofing book, devoted to the most fascinating problem which Europe
will hove to deal with in the coming decades. The President of the European Investment
Bonk, Sir Brian Unwin, who provided α detailed introducfion to the book, rightly
observes: "In order to ensure that the book is not quickly outdoted by the rapid evolution
of polifics, the authors hove concentrated on issues of lasting interest" (page XI).
My hnol impression is that this book is economics at its best.
H.W.J. Bosman,
em. Professor Tilburg University, Netherlands
EIB Papers Volumel Noi 1996 145
J o i n t S t a t e m e n t i s s u e d b y t h e E u r o p e a n C o m m i s s i o n
a n d t h e EIB o n J u l y 3 1 , 1 9 9 6
Since tfie conclusions of the European Council in M a d r i d in December 1995, the
prospectus of all new debt instruments issued by the European Community, the
European Cool and Steel Community, Euratom and the European Investment Bank have
contained α clause confirming the principle of one to one continuity between the ECU
and the Euro. From the start of monetary union, on 1 January 1999, all interest payments
and repayments of the principal will be mode in the single European currency, the Euro.
The substitution will be at α rate of one for one in line with the Treaty on European
Union (the Maastricht Treaty), in particular Article 109(4].
The European Commission confirms that this same principle will apply to all ECU-den
ominoted loon instruments issued by the European institutions, including those dating
from before the Madr id European Council. All outstanding loon instruments will there
fore be treated in the same way at the start of monetary union.
Thus the Commission and the European Investment Bank have the same policy.
The Commission intends shortly to table proposals for Council legislation on the legal
status of the Euro containing specific provisions confirming the general principle of one
to one conversion for outstanding loan instruments denominated in ECU;
The total value of outstanding ECU-denominoted instruments managed by the European
Commission with maturity after I January 1999 is ECU 2 0 1 0 million. For the EIB the
equivalent figure is ECU 6 0 0 0 million. This situation is clearly independent of future
issues in ECU between now and December 1998.